Method and system for dynamic compensation of false touch of vehicle seat micro switch

Abstract

The present application relates to the technical field of vehicle seat control, and specifically discloses a dynamic compensation method and system for preventing false touch of a micro switch of a vehicle seat, the method comprising: synchronously collecting three-axis space acceleration signals and loop real-time current signals corresponding to each micro switch through a hardware triggering mode; extracting features from the three-axis space acceleration signals to obtain real-time frequency spectrum features, matching the real-time frequency spectrum features with pre-stored mechanical resonance fingerprints to generate a resonance flag bit. In the actual operation of a vehicle, the technical solution of the present application synchronously acquires space acceleration and loop current signals, and uses the comparison between frequency spectrum features and pre-stored mechanical resonance fingerprints to determine the resonance state of the micro switch. The system divides scenarios in combination with multi-dimensional information such as road condition types, service life of the micro switch, and online estimated contact resistance, and accordingly performs layered dynamic compensation calculation under different working conditions to separate interference components caused by vibration from real-time current.

Description

Technical Field

[0001] This invention relates to the field of vehicle seat control technology, and in particular to a dynamic compensation method and system for preventing accidental touches of micro switches for vehicle seats. Background Technology

[0002] The adjustment function of a car seat is usually achieved by a microswitch installed inside. During vehicle operation, continuous mechanical vibrations are generated. Existing methods to prevent accidental activation of microswitches typically collect the current signal of the switch circuit and use a fixed current judgment threshold or fixed filtering parameters to filter out electrical signal fluctuations caused by normal vehicle bumps.

[0003] When the vehicle's operating conditions cause the external vibration frequency to approach the switch's own mechanical natural frequency, the internal spring of the microswitch will generate mechanical resonance, causing the contacts to bounce at high frequency. The current fluctuation characteristics caused by this resonance have a high degree of overlap with the electrical signal characteristics generated by manually pressing the switch. Existing fixed parameter processing methods have difficulty distinguishing between manual operation signals and mechanical resonance interference signals. Furthermore, as the microswitch is used for longer periods, the contact resistance of its internal contacts will change, which will cause the originally set fixed judgment parameters to fail, leading to false start-up of the seat adjustment motor or failure to respond to actual pressing operations. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in related technologies. Therefore, the objective of this invention is to propose a dynamic compensation method and system for preventing accidental activation of microswitches in vehicle seats, thereby improving the operational stability of seat adjustment systems.

[0005] To achieve the above objectives, a first aspect of the present invention provides a dynamic compensation method for preventing accidental touches of a micro switch for a vehicle seat, comprising:

[0006] The three-axis spatial acceleration signal and the real-time loop current signal corresponding to each micro switch are synchronously acquired through hardware triggering.

[0007] The real-time spectral features are obtained by extracting features from the triaxial spatial acceleration signal. The real-time spectral features are then matched with the pre-stored mechanical resonance fingerprint to generate a resonance flag.

[0008] Obtain the road condition type of the vehicle where the micro switch is located and the service life level of the micro switch, and calculate the online estimated contact resistance corresponding to the real-time current signal of the circuit;

[0009] By combining the road condition type, the service life level of the micro switch, the online estimated contact resistance, and the resonance flag, a four-dimensional scene classification result is generated.

[0010] Based on the four-dimensional scene classification results, perform hierarchical dynamic compensation calculation to obtain the preliminary state and accurate pure actual current.

[0011] Based on the precise and pure actual current, construct the state confidence vector of all microswitches, and combine it with the preset multi-switch state correlation matrix to establish a multi-objective optimization objective function, and solve to obtain the global optimal switch state combination.

[0012] The globally optimal switch state combination is output to the vehicle controller, and a graded fault handling process is executed in conjunction with the three-axis spatial acceleration signal, the real-time loop current signal and the resonance flag.

[0013] To achieve the above objectives, a second aspect of the present invention provides a dynamic compensation system for preventing accidental touches of a micro switch for a vehicle seat, comprising:

[0014] The synchronous acquisition module is used to synchronously acquire the three-axis spatial acceleration signal and the real-time current signal of the loop corresponding to each micro switch through hardware triggering.

[0015] The feature extraction and matching module is used to extract features from the triaxial spatial acceleration signal to obtain real-time spectral features, and match the real-time spectral features with the pre-stored mechanical resonance fingerprint to generate a resonance flag bit.

[0016] The parameter acquisition and scene classification module is used to acquire the road condition type of the vehicle where the micro switch is located and the service life level of the micro switch, and calculate the online estimated contact resistance corresponding to the real-time current signal of the circuit; and combine the road condition type, the service life level of the micro switch, the online estimated contact resistance and the resonance flag to generate a four-dimensional scene classification result.

[0017] The hierarchical compensation solution module is used to perform hierarchical dynamic compensation solution based on the four-dimensional scene classification results to obtain the preliminary state and the accurate pure actual current.

[0018] The joint state determination module is used to construct the state confidence vector of all microswitches based on the accurate pure actual current, and to establish a multi-objective optimization objective function by combining the preset multi-switch state correlation matrix, and to solve for the global optimal switch state combination.

[0019] The output and fault handling module is used to output the globally optimal switch state combination to the vehicle controller, and to perform a graded fault handling process in conjunction with the three-axis spatial acceleration signal, the real-time loop current signal and the resonance flag.

[0020] To achieve the above objectives, a third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory. When the computer program is executed by the processor, it implements the above-described method for dynamic compensation to prevent accidental touches of a vehicle seat micro switch.

[0021] The technical solution of this invention determines the resonance state of a microswitch by synchronously acquiring spatial acceleration and loop current signals during actual vehicle operation and comparing their spectral characteristics with pre-stored mechanical resonance fingerprints. The system combines multi-dimensional information such as road condition type, microswitch lifespan, and online estimated contact resistance to classify scenarios and performs layered dynamic compensation calculations under different operating conditions, removing vibration-induced interference components from the real-time current. This solution adapts to changes in the physical characteristics of the microswitch throughout its service life and, by combining multiple switch state correlation matrices for joint solution, reduces the probability of misjudgment due to local mechanical resonance or aging of contact components. This ensures that the final output switch state accurately reflects the operator's true intentions, improving the operational stability of the seat adjustment system. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the dynamic compensation method for preventing accidental touches of the vehicle seat micro switch provided by the present invention.

[0023] Figure 2 This is a time-frequency distribution diagram of the triaxial acceleration signal under mechanical resonance conditions in the dynamic compensation method for preventing accidental touch of the vehicle seat micro switch provided by the present invention.

[0024] Figure 3 This is a Kalman recursive convergence curve of the resonant fingerprint aging correction coefficient in the dynamic compensation method for preventing accidental touch of the vehicle seat micro switch provided by the present invention.

[0025] Figure 4 This is a nonlinear response surface diagram of contact displacement and resonant interference current in the anti-misoperation dynamic compensation method for micro switches of vehicle seats provided by the present invention.

[0026] Figure 5 This is a timing diagram of the spatial acceleration Z-axis low-frequency gravity baseline separation in the dynamic compensation method for preventing accidental touch of the vehicle seat micro switch provided by the present invention;

[0027] Figure 6 This is a coupled response surface diagram of occupant load, sitting posture coefficient and transient resonant frequency in the dynamic compensation method for preventing accidental touch of the vehicle seat micro switch provided by the present invention.

[0028] Figure 7This is a schematic diagram of the coherence of the spatial phase-locked value (PLV) on the complex plane unit circle in the dynamic compensation method for preventing accidental touch of the vehicle seat micro switch provided by the present invention.

[0029] Figure 8 This is a schematic diagram illustrating the implementation of the anti-accidental touch dynamic compensation system for the micro switch of the vehicle seat provided by the present invention;

[0030] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0031] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0032] The following description, with reference to the accompanying drawings, describes a dynamic compensation system, method, and electronic device for preventing accidental touches of a vehicle seat micro switch according to embodiments of the present invention.

[0033] Example 1:

[0034] In modern vehicles, especially those with multi-directional electric adjustment capabilities, numerous microswitches are integrated within the seats. These microswitches, serving as direct inputs for human-machine interaction, directly determine the correctness of the seat adjustment motor's operation based on the accuracy of their electrical states.

[0035] However, during vehicle operation, the broadband vibrations transmitted from the chassis to the seat frame often couple to the reed structure of the microswitch. This embodiment provides a dynamic compensation method for preventing accidental activation of vehicle seat microswitches. This method relies on the coordinated cooperation of the underlying hardware acquisition architecture and the upper-level multi-dimensional data processing algorithm to extract interference features caused by mechanical vibration from multi-source heterogeneous signals.

[0036] like Figure 1 As shown, the method for preventing accidental activation of the micro switch for the vehicle seat includes the following steps:

[0037] Phase 1: High-precision spatiotemporal synchronous acquisition and preprocessing of underlying signals.

[0038] Specifically, the execution steps are as follows: The three-axis spatial acceleration signal and the real-time loop current signal corresponding to each microswitch are synchronously acquired via hardware triggering. .

[0039] It is important to note that traditional electrical signal acquisition often employs software polling to read sensor data. However, due to task scheduling jitter and interrupt response delays in the operating system of the vehicle-mounted microcontroller, software polling can lead to a millisecond-level random phase difference between the acceleration signal and the current signal. In dynamic compensation calculations, this phase error is amplified, causing compensation failure. Therefore, the hardware triggering method in this embodiment involves configuring an independent global universal timer within the microcontroller. This timer generates a pulse-width modulation signal or trigger pulse with a fixed period. This trigger pulse directly drives the analog-to-digital converter for current sampling and simultaneously drives the serial peripheral interface to read the data register of the microelectromechanical system's accelerometer, thereby ensuring that the three-axis spatial acceleration signal and the real-time loop current signal are synchronized. Achieve microsecond-level strict alignment on the timeline.

[0040] For example, the triaxial spatial acceleration signal is acquired by a triaxial microelectromechanical system (MEMS) sensor array arranged on rigid nodes of the seat frame near the microswitch. This signal includes an X-axis component along the vehicle's longitudinal direction, a Y-axis component along the vehicle's transverse direction, and a Z-axis component along the vehicle's vertical direction. Real-time loop current signal. It is a digital sequence obtained by an analog-to-digital converter through a high-precision shunt resistor connected in series in the microswitch signal circuit and adjusted by a differential amplifier circuit. It reflects the mixed current state of the microswitch circuit at the current moment, which contains the real trigger signal and the mechanical vibration interference signal.

[0041] The second stage: extraction of frequency domain features and temperature-corrected mechanical resonance matching.

[0042] Specifically, the execution steps are as follows: feature extraction is performed on the three-axis spatial acceleration signal to obtain real-time spectrum features, and the real-time spectrum features are matched with the pre-stored mechanical resonance fingerprint to generate resonance flag bits.

[0043] In actual vehicle operating environments, microswitches do not produce false triggers at all vibration frequencies. Resonance typically occurs only when the external excitation frequency approaches the inherent mechanical frequency of the cantilever beam of the internal metal contacts of the microswitch, leading to drastic fluctuations in contact pressure or even contact disconnection. To accurately identify this operating condition, the feature extraction and matching process in this embodiment specifically includes the following steps:

[0044] First, the three-axis spatial acceleration signals are vector-synthesized to obtain a composite acceleration signal. Since the microswitch may have a spatial tilt angle during installation inside the seat, acceleration along a single axis cannot fully reflect the excitation energy experienced by the switch. Therefore, the Euclidean norm algorithm is used to calculate the square root of the sum of the squares of the acceleration components in the three orthogonal directions (X-axis, Y-axis, and Z-axis), generating a scalar sequence, i.e., the composite acceleration signal, to characterize the comprehensive spatial excitation force experienced by the switch.

[0045] Subsequently, a sliding-window short-time Fourier transform is performed on the synthesized acceleration signal to extract the real-time peak frequency and peak amplitude. Conventional fast Fourier transforms lack time resolution and cannot pinpoint the exact moment the vibration occurs. By introducing a Hanning or Hamming window as a sliding time window, the synthesized acceleration signal is segmented and windowed before Fourier transform, allowing the acquisition of the signal's time-frequency distribution matrix. Within the current time slice of this matrix, the highest peak in the amplitude spectrum is searched; the frequency coordinate corresponding to this peak is the real-time peak frequency, and the corresponding amplitude is the peak amplitude.

[0046] like Figure 2 This figure illustrates the time-frequency distribution of the triaxial acceleration signal when a microswitch encounters mechanical resonance during vehicle operation. The horizontal axis represents the time series of the system operation in seconds; the vertical axis represents the distribution frequency of the vibration signal in Hertz; and the color bars on the side of the figure map the transient energy of the composite acceleration in gravitational acceleration, with dark blue indicating a low-energy state and dark red indicating a high-energy accumulation state.

[0047] As can be seen from the data in the figure, within the overall observation interval of 0 to 0.5 seconds, there is a continuous light blue distribution band in the extremely low frequency region from 0 Hz to 100 Hz. This reflects the background low-frequency vibration component caused by road bumps during normal vehicle driving. The excitation energy of this component is low and does not meet the conditions for triggering contact separation.

[0048] Within a time interval of 0.2 to 0.3 seconds, a significant dark red high-energy accumulation patch suddenly appeared at the 600 Hz frequency coordinate position. This waveform change and color jump objectively reflect that the external broadband excitation force at this moment stimulated the inherent mode of the cantilever beam of the internal metal contact of the microswitch, triggering a mechanical resonance phenomenon.

[0049] The system extracts the time-frequency matrix by performing a sliding window short-time Fourier transform on the triaxial spatial acceleration signal, enabling direct addressing of the 600 Hz real-time peak frequency and corresponding peak amplitude of the dark red patch. When the 600 Hz frequency falls within the inherent resonant frequency matching window corrected for the current ambient temperature, and the amplitude corresponding to the dark red patch reaches the first amplitude threshold to overcome the static preload of the contact, the system determines that the microswitch has entered the resonant operating condition, thus setting the resonant flag to an active state.

[0050] This chart provides intuitive data evidence of the structural abrupt changes in signal energy in both spatiotemporal dimensions, demonstrating the physical process and practical effect of the system stripping specific high-frequency local resonance features from a mixed background of conventional mechanical vibrations.

[0051] It is also important to note that the elastic modulus of the cantilever beam material of a microswitch (such as beryllium bronze or phosphor bronze) changes with ambient temperature, which directly results in its mechanical natural frequency not being a fixed value. Therefore, it is necessary to obtain the current ambient temperature and calculate the corrected natural resonant frequency based on it.

[0052] For example, the formula for calculating the corrected natural resonant frequency is as follows:

[0053] ;

[0054] In the formula, This indicates the corrected natural resonant frequency. This indicates the inherent resonant frequency of the microswitch as specified at the factory. This represents the first-order temperature correction factor. This represents the second-order temperature correction factor. Indicates the current ambient temperature. This indicates the preset reference ambient temperature. The first-order and second-order temperature correction coefficients are obtained by full-temperature-range scanning calibration in a thermodynamic environment before the switching components leave the factory, and are used to characterize the nonlinear mapping relationship between the material's elastic modulus and temperature.

[0055] After obtaining the correction parameters, the resonance matching degree between the real-time peak frequency and the corrected natural resonant frequency is calculated. This matching degree can be obtained by calculating the absolute value of the frequency difference between the two and substituting it into a Gaussian decay function or its reciprocal function, which is used to quantify how close the current external vibration frequency is to the switch's natural frequency.

[0056] When the resonance matching degree is greater than or equal to the first matching threshold and the peak amplitude is greater than or equal to the first amplitude threshold, the corresponding microswitch is determined to have entered the resonance condition, and the resonance flag is set to an active state; otherwise, the resonance flag is set to an inactive state. The first matching threshold sets the tolerance boundary for frequency overlap, and the first amplitude threshold sets the physical boundary that the excitation energy must reach to overcome the static preload of the contacts. Only when the frequencies are aligned and the energy is sufficient is the resonance flag set to an active state (e.g., logic value 1), which greatly suppresses system oversensitivity caused by low-energy background noise.

[0057] The third stage: the aging self-calibration mechanism of mechanical resonance fingerprints.

[0058] Besides being affected by temperature, mechanical resonant fingerprints also experience long-term, irreversible aging drift in their inherent resonant frequency due to mechanical fatigue caused by prolonged use of microswitches. This is caused by the release of internal stress in the metal structure and microscopic wear of the geometric dimensions. Therefore, the mechanical resonant fingerprint in this embodiment includes an aging correction coefficient, and the method further includes a step of self-calibrating the aging correction coefficient.

[0059] Specifically, sample data under resonance conditions are collected at preset intervals, and the aging correction coefficient is updated using the recursive least squares method. The calculation formula is as follows:

[0060] ;

[0061] In the formula, This represents the updated aging correction factor. This represents the current aging correction factor. Indicates Kalman gain, Indicates the first The real-time peak frequency extracted when the system first enters the resonance condition. This indicates the inherent resonant frequency of the microswitch as specified at the factory. This indicates the current service life of the corresponding microswitch.

[0062] It's also important to note that, compared to the traditional moving average algorithm, the recursive least squares method does not require storing massive amounts of historical data samples in memory. Each time a resonance condition actually occurs during vehicle operation, the system treats the measured real-time peak frequency as the actual physical observation value of the switch, uses Kalman gain control to adjust the step size, dynamically eliminates observation noise, and gradually approximates the true aging rate. Through this step, the system is endowed with the ability to dynamically update the pre-stored fingerprint along with the microswitch's lifespan, avoiding the failure of the anti-mistouch algorithm due to an outdated fingerprint database after several years of vehicle service.

[0063] like Figure 3The Kalman recursive convergence state of the aging correction coefficient for the mechanical resonance fingerprint is shown. The horizontal axis in the figure represents the number of iterations of the system collecting resonance condition sample data, in units of iterations; the vertical axis represents the aging correction coefficient of the microswitch's inherent resonant frequency, in units of dimensionless.

[0064] The graph contains a horizontal red dashed line and a blue solid line that shows a dynamic upward trend. The red dashed line represents the actual aging coefficient of the microswitch contact structure after long-term use, which is stable at 0.015. The blue solid line represents the recursive estimate coefficient calculated after the system extracts the real-time peak frequency each time under resonant conditions.

[0065] As can be seen from the curve changes and waveform transformations in the figure, in the initial stage of system operation, from 0 to 20 iterations, due to the small accumulation of effective samples and the large interference of external measurement noise, the blue solid line shows a certain degree of random fluctuation and is in the lower data range.

[0066] As the number of sample data collected according to the preset cycle continues to increase, when the number of iterations reaches about 40, the blue solid line shows a clear convergence trend, its value rapidly approaches the red dashed line, and the fluctuation amplitude narrows significantly.

[0067] When the number of iterations reaches 80 to 100, the fluctuations of the blue solid line basically subside and stabilize at the data baseline of 0.015, tending to coincide with the red dashed line representing the real physical state.

[0068] This data chart objectively reflects the actual operating effect of the system by using recursive least squares method and Kalman gain dynamic control to correct the step size. It proves that the aging self-calibration mechanism can gradually filter out observation noise and approach the real mechanical fatigue aging rate without retaining massive amounts of historical data. It realizes the function of synchronously updating the pre-stored mechanical resonance fingerprint with the service life of the micro switch, and improves the long-term operating stability of the anti-mistouch algorithm.

[0069] Phase 4: Four-dimensional scene classification and online resistance estimation based on multi-dimensional feature fusion.

[0070] The changes in the electrical characteristics of a microswitch are not caused by a single factor, but are influenced by a combination of environmental factors, historical conditions, and real-time operation. Specifically, the steps are as follows: obtain the road condition type of the vehicle where the microswitch is located and the service life rating of the microswitch, and calculate the real-time current signal of the circuit. The corresponding online estimated contact resistance. For example, road condition types can be classified by performing time-domain roughness statistics on the triaxial spatial acceleration signal (such as calculating the root mean square value, kurtosis, crest factor, etc.), for example, smooth asphalt roads, cobblestone roads, continuous speed bumps, etc. The service life rating of the microswitch is obtained by mapping the operating time and cumulative number of presses recorded in the vehicle's onboard memory.

[0071] Meanwhile, due to arc erosion on the switch contact surface during repeated closing and opening, the surface oxide layer thickens, significantly increasing its static contact resistance. To obtain this parameter, the steps of acquiring the switch circuit supply voltage and the inherent resistance of the circuit of the microswitch are performed. Based on the switch circuit supply voltage, the inherent resistance of the circuit, and the precise pure actual current when the microswitch is closed... The steady-state average value is used to calculate the online estimated contact resistance, and the calculation formula is as follows:

[0072] ;

[0073] In the formula, This represents the online estimated contact resistance. Indicates the power supply voltage of the switching circuit. This indicates the precise and pure actual current when the microswitch is closed. The steady-state average value, This represents the inherent resistance of the circuit. The steady-state average value is extracted here to avoid the transient mechanical bounce at the moment of pressing and the inductive surge current that may exist in the subsequent load circuit, ensuring the accuracy of the reference physical quantity for resistance estimation.

[0074] Subsequently, by combining road condition type, microswitch service life level, online estimated contact resistance, and resonance flag, a four-dimensional scene classification result is generated. This classification result is essentially a multi-dimensional mapping matrix. For example, one classification result might be characterized as: "Smooth asphalt road - low service life - low contact resistance - no resonance," while another classification result might be characterized as: "Cobblestone road - high service life - high contact resistance - resonance occurs." Through this high-dimensional scene segmentation, the system can accurately match the most suitable compensation algorithm for switches in different physical states, avoiding the decline in the applicability of a single model throughout its entire life cycle.

[0075] Phase 5: Layered dynamic compensation solution for specific scenarios.

[0076] After obtaining the scene classification, the system enters the core signal purification stage. Specifically, based on the four-dimensional scene classification results, hierarchical dynamic compensation calculations are performed to obtain the preliminary state and the accurate and pure actual current. The basis for the layered solution here lies mainly in the state of the resonance flag, because the mechanical model of the switch contact exhibits completely different mathematical laws in the resonant and non-resonant states.

[0077] Optionally, when the resonance flag is invalid, it indicates that the current vibration is mainly low-frequency or non-periodic impact, the contacts have not separated, but the contact pressure fluctuates dynamically under the action of vibration acceleration, which will cause changes in the area of ​​the microscopic conductive spots on the contact surface. In this case, the compensation impedance is calculated using the dedicated cross-domain transfer function model for non-resonant scenarios:

[0078] ;

[0079] In the formula, Indicates the compensation impedance. This represents the transient energy envelope extracted from the triaxial spatial acceleration signal. , , Represents the nonlinear coefficients. This represents the contact resistance correction factor. This represents the online estimated contact resistance. The formula utilizes the Taylor expansion approach, employing a high-order polynomial to fit the nonlinear modulation effect of the non-resonant impact energy on the contact impedance.

[0080] Subsequently, based on the compensation impedance Calculate the non-resonant interference current component based on the supply voltage of the switching circuit (according to Ohm's law), and obtain the real-time current signal from the circuit. By subtracting the non-resonant interference current component, the precise and pure actual current, after filtering out background low-frequency vibration noise, can be obtained. .

[0081] Optionally, when the resonance flag is active, it indicates that the contact is in a critical or separated state of high-frequency bouncing. At this time, the impedance model is no longer continuous, and a micro-displacement model needs to be introduced. Since the supply voltage and inherent line resistance have already been obtained in the previous non-resonant calculation and contact resistance estimation steps, the step of obtaining the insulation resistance of the microswitch in the open state is performed first here. The resonant interference current component is calculated using the dedicated dynamic compensation model for resonance:

[0082] ;

[0083] In the formula, This represents the resonant interference current component. Indicates the power supply voltage of the switching circuit. Indicates the inherent resistance of the circuit. This indicates the insulation resistance when the microswitch is in the off state. The nonlinear coefficient characterizing the abrupt change rate of contact separation impedance, This indicates the contact threshold displacement. This represents the relative displacement of the contact point at resonance, calculated based on the corrected resonant gain coefficient and the corrected resonant damping ratio.

[0084] It's also important to note that the denominator in this formula contains a nested Sigmoid activation function structure. Its physical meaning lies in: when the calculated relative displacement of the resonant contact point... Much smaller than the contact threshold When the displacement approaches or exceeds the threshold, the function term approaches 0, and the switch exhibits an insulation resistance value that is approximately open. When the displacement approaches or exceeds the threshold, the function output changes drastically, simulating the step drop in impedance at the instant the contacts close. Through this cross-domain mapping formula from dynamics to electrical engineering, the system can accurately reconstruct the high-frequency distortion current induced by mechanical resonance. Similarly, it can also reconstruct the real-time current signal from the circuit. Subtracting the resonant interference current component This allows us to obtain precise and pure actual current under harsh resonant conditions. .

[0085] like Figure 4 This figure illustrates the nonlinear response surface characteristics of contact displacement and resonant interference current under resonant conditions. The horizontal axis represents the relative displacement of the contacts in millimeters; the vertical axis represents the impedance abrupt change rate at contact separation in dimensionless units; and the vertical axis represents the calculated resonant interference current in amperes. The color distribution on the surface smoothly transitions from dark blue to dark red, corresponding to the order of magnitude change in current value.

[0086] Based on the data and surface geometry in the figure, it can be seen that when the relative displacement of the contact is in the range of 0 to 1.0 mm, the resonant interference current remains at an extremely low level of close to 0.005 amperes, and the surface appears as a flat dark blue area. This reflects that the micro switch still maintains a relatively stable high impedance state under slight mechanical vibration.

[0087] When the relative displacement of the contact points crosses the 1.0 mm contact threshold node, the response surface undergoes a sharp, step-like morphological transformation. Influenced by the modulation of the impedance mutation rate parameter on the vertical axis, the current value rapidly increases along the vertical axis and eventually saturates in the high range of nearly 10 amperes, forming a deep red high-current plateau.

[0088] Specifically, when the impedance abrupt change rate increases from 5 to 30, the transition slope of the surface at the 1.0 mm displacement node increases significantly, and the transition band of the current jump narrows accordingly.

[0089] This three-dimensional surface chart intuitively demonstrates the cross-domain mapping relationship between dynamic displacement and electrical current of the resonance-specific dynamic compensation model in the specific implementation, confirming that the nested model can accurately reconstruct the rapid change process of high-frequency distortion current caused by mechanical resonance, thus providing reliable data derivation support for the system to accurately deduct the resonance interference component from the real-time current signal.

[0090] Phase 6: Resonance Enhancement Judgment Logic.

[0091] To prevent sporadic electrical and electromagnetic interference, such as the instantaneous activation of high-power radio frequency equipment inside the vehicle, from being mistakenly treated as mechanical resonance, a precise and pure actual current is used as the basis for... Before constructing the state confidence vector, the method also includes a strict validation mechanism:

[0092] When the resonance flag is active, the threshold for determining the state of the corresponding microswitch is updated from a first duration to a second duration, where the second duration is longer than the first duration. For example, the system originally only needed a high level for 3 milliseconds (the first duration) to determine that the switch was pressed and closed. However, when resonance is detected, the system's confidence in the signal decreases, requiring the high level to last for 10 milliseconds (the second duration) before proceeding to the next determination. This is equivalent to adding a filtering window in the time dimension.

[0093] At the same time, calculate the accurate and pure actual current. Total Harmonic Distortion (THD) is the ratio of the energy of all higher harmonics to the fundamental frequency energy. The current signal generated by a genuine manual pressing action has a relatively smooth envelope and low THD; however, even after compensation, the residual current from mechanical resonance still carries a high proportion of high-frequency harmonics. Based on the distribution of THD within a preset range defined by a first distortion threshold and a second distortion threshold, the signal is determined to be either a genuine trigger signal or a resonance interference signal, where the second distortion threshold is greater than the first distortion threshold. If the distortion is lower than the first distortion threshold, it tends to be a genuine trigger; if it is higher than the second distortion threshold, it tends to be resonance interference.

[0094] Furthermore, a spatial consistency check is introduced. The resonance flags of three adjacent microswitches in space are detected. If the current microswitch does indeed resonate, it is usually due to local structural features of its mounting location or individual wear. If the resonance flags of adjacent microswitches are all invalid, it indicates that the vibration is a local phenomenon, which is consistent with the drift pattern of the microswitch's own characteristics. The confidence level of the current microswitch's state is then further confirmed by combining the total harmonic distortion (THD) assessment results. If all adjacent switches report resonance simultaneously, it is highly likely that the entire vehicle has experienced extreme forced vibration, in which case the confidence level will be significantly weakened.

[0095] Phase 7: Solving the global switch state for multi-objective optimization.

[0096] Traditional vehicle control logic often determines the state of each microswitch in isolation. However, this ignores ergonomic constraints. For example, when a passenger adjusts the seat back to recline, the mechanical constraints of their hand bones make it extremely difficult to simultaneously trigger the switch that moves the seat cushion forward. Therefore, the execution steps are based on precise, pure, actual current. Construct the state confidence vector of all microswitches, establish a multi-objective optimization objective function in combination with the preset multi-switch state correlation matrix, and solve for the globally optimal switch state combination.

[0097] For example, the expression for the multi-objective optimization objective function is as follows:

[0098] ;

[0099] In the formula, Represents the objective function value. Indicates the first The weight of each microswitch This represents the confidence level in the state confidence vector. Indicates the current time. A state mapping value for a micro switch, which includes two discrete states: open and closed (usually defined as 1 for closed and 0 for open). Indicates the current time. The state mapping value of a micro switch.

[0100] In the second half of the equation: This represents an element of a preset multi-switch state correlation matrix, which records the extreme probabilities of different switches being triggered simultaneously. This represents the consistency penalty coefficient. If the algorithm's output state combinations contain operation combinations that are contradictory to human behavior or have extremely low probability, the penalty value of the second term will increase significantly.

[0101] Represents the continuity penalty coefficient. Indicates the previous moment. The state mapping value of a micro switch. The human action of pressing the switch must be continuous, and it is impossible to continuously perform the "on-off-on-off" toggle operation within 10 milliseconds. Therefore, a penalty term for the state change rate is introduced to suppress high-frequency jitter.

[0102] Represents the penalty coefficient for the resonant state. This represents the resonant flag mapping value of the corresponding microswitch. If a switch is under strong resonant interference and is determined to be triggered, the system will impose a penalty on the total cost function, forcing the algorithm to treat the data of that channel more cautiously. This objective function is then evaluated using a sequential quadratic programming algorithm or a heuristic search algorithm. Perform a minimization solution and output the result. The vector is the globally optimal combination of switching states that conforms to physical laws, ergonomics, and electrical reliability.

[0103] Phase 8: Safety classification and fault handling mechanism.

[0104] After completing the state calculation, the following steps are executed: The globally optimal combination of switch states is output to the vehicle controller as a direct control command for the seat adjustment motor drive axle. Simultaneously, to ensure the safety of the system throughout its entire lifecycle, three-axis spatial acceleration signals and real-time loop current signals are incorporated. A rigorous graded fault handling process is implemented in conjunction with the resonance flag, specifically including:

[0105] When the online estimated contact resistance increases by more than a first preset multiple of its initial value, such as 2.5 times the factory initial value, it indicates that the internal contacts of the switch have undergone substantial and severe oxidation or surface plating peeling. At this time, a level one fault is triggered. The system does not shut down directly, but automatically adjusts the parameters of the layered dynamic compensation calculation (such as increasing the correction coefficient in the aforementioned compensation impedance polynomial) to compensate for signal attenuation, and records the fault log to non-volatile memory so that it can be read through the diagnostic interface during vehicle maintenance.

[0106] When the number of intermittent contact failures detected by the microswitch within a unit of time exceeds a preset threshold, it indicates that the mechanical structure may have become fatigued and loose, and the contacts cannot maintain stable electrical contact under normal pressing pressure. This triggers a level-two fault. The system not only logs the information but also actively increases the state confidence threshold in the multi-objective optimization objective function. This means that subsequent operations require more stable and significant electrical signals to be accepted by the system, preventing random malfunctions caused by loose contacts.

[0107] When a microswitch contact is determined to be stuck (e.g., a continuous high current is detected in the circuit, the acceleration signal does not show any impact vibration, and the user cancels the pressing action command), the contacts physically stick together due to the high temperature of the electric arc. This is an extremely dangerous runaway fault that will trigger a level 3 fault. The system forcibly cuts off the power supply to the corresponding seat adjustment motor via bus commands or hard-wired relays to prevent the motor from stalling for an extended period, which could lead to damage to the seat's mechanical structure or thermal runaway and fire of the wiring harness.

[0108] When the amplitude of the three-axis spatial acceleration signal is less than the preset static judgment threshold, i.e., the vehicle is in a stable state of absolute stillness and no occupant movement, and the system continuously outputs a valid resonance flag, this violates the laws of physics and indicates that the internal hardware sensors of the system are damaged (such as the levitation mass block inside the MEMS being stuck, or a short circuit oscillation occurring in the ADC input channel). At this time, a level four fault is triggered, which directly triggers a safety-level alarm through the vehicle controller, illuminates the instrument panel fault light, and prohibits the system from executing any automated adjustment procedures.

[0109] In general, existing technologies for preventing accidental touches of automotive microswitches typically employ fixed debounce delay filtering or simple level voltage comparator architectures. These static and isolated methods are prone to misinterpreting the spike currents caused by mechanical resonance as user intent when faced with complex road surface vibrations. Furthermore, they cannot withstand the aging of contact resistance and mechanical resonance frequency drift caused by long-term use of switching components.

[0110] The technical solution provided in this embodiment establishes a cross-domain dynamic compensation model covering mechanical kinematics and electrical response through high-precision, strictly synchronized acquisition of multiple physical quantities (acceleration and current). By calculating the relative displacement of contacts under vibration and impact in real time and establishing Taylor impedance polynomials and nonlinear excitation equations, the system achieves the goal of extracting accurate and pure signals from contaminated real-time current. The introduced aging self-calibration algorithm and four-dimensional scene classification logic enable the system to have adaptive growth capabilities. The idea of ​​joint optimization of multiple switch states breaks through the limitations of previous single-point methods, effectively avoiding the risk of seat misadjustment caused by harsh working conditions and component aging, and ensuring the stable execution of the occupant's human-machine interaction experience and driving safety control logic.

[0111] Example 2:

[0112] Specifically, before matching the real-time spectral features with the pre-stored mechanical resonance fingerprint, the method of this embodiment also includes a dynamic fingerprint correction step for frequency shifts caused by occupant mass and posture coupling.

[0113] In Example 1, the system primarily performs frequency domain window matching using short-time Fourier transform based on the inherent resonant frequency of the microswitch after factory calibration or temperature correction. However, in real-world vehicle physics applications, the vehicle seat and its associated microswitch are not isolated, constant-mass rigid systems. When an occupant sits down, their body mass acts as a large additional mass, directly coupling into the mechanical vibration system of the seat frame. According to the fundamental principles of classical structural dynamics, the inherent resonant frequency of a composite vibration system is inversely proportional to the square root of its equivalent modal mass. Therefore, the addition of the occupant's mass significantly alters the "mass-spring" model of the entire seat's local system, causing a noticeable nonlinear shift in the actual resonant frequency (typically exhibiting a trend towards lower frequencies).

[0114] For example, the equivalent additional mass experienced by the seat side frame differs significantly depending on whether the seat carries a 40kg child or a 100kg adult. If the anti-mistouch system still rigidly uses the pre-stored mechanical resonance fingerprint from the unloaded state as a fixed matching benchmark, the actual resonance peak frequency excited by the seat frame will deviate significantly from the preset frequency domain matching search window when a heavily loaded occupant compresses the seat and encounters severe road conditions. This will cause the resonance matching degree calculated by the algorithm to be lower than the set matching threshold, resulting in severe false negatives and preventing the subsequent resonance-specific dynamic compensation model from being correctly triggered, thus rendering the anti-mistouch mechanism ineffective.

[0115] To address this technical problem caused by the dynamic time-varying nature of physical boundary conditions, this embodiment introduces a spatial dimensionality reduction and dynamic coupling correction mechanism based on existing triaxial spatial acceleration signals, without adding additional physical weighing sensors or thin-film pressure sensors.

[0116] Specifically, the Z-axis component of the triaxial spatial acceleration signal is low-pass filtered to extract the extremely low-frequency gravity acceleration baseline, and the equivalent static load of the current occupants is calculated based on the extremely low-frequency gravity acceleration baseline. The steps.

[0117] For example, a microelectromechanical system (MEMS) accelerometer, while outputting a high-frequency dynamic alternating vibration signal, also possesses a stable DC response capability, enabling it to continuously sense the static projection components of the Earth's gravitational field along its various sensing axes. When the vehicle is traveling on a level or near-level road surface, the Z-axis of the sensor coordinate system (i.e., the vertical direction perpendicular to the vehicle chassis) primarily bears the continuous... Standard gravitational acceleration. Since the transient bumps caused by uneven road surfaces, suspension rebound, and high-frequency excitation signals from the engine are usually distributed in a higher frequency band, this embodiment uses a digital low-pass filter with an extremely low cutoff frequency to process the original Z-axis component sequence.

[0118] Optionally, the digital low-pass filter can be a Butterworth low-pass filter or a Chebyshev low-pass filter with adjustable order. To effectively filter out signal glitches caused by dynamic impacts from the road surface, the filter's cutoff frequency is typically set in an extremely low frequency range of a few tenths of a hertz, such as 0.1 Hz to 0.5 Hz. Through this digital signal processing step, the system can smoothly strip away the high-frequency AC dynamic acceleration components in the Z-axis component, thereby outputting a smooth, continuous, and extremely low-frequency gravitational acceleration baseline that represents only the steady state of gravity.

[0119] like Figure 5 This figure illustrates the temporal evolution of low-frequency gravity baseline separation along the Z-axis of spatial acceleration. The horizontal axis represents the system's operating time in seconds; the vertical axis represents the Z-axis acceleration sensed by the sensor, in units of gravitational acceleration.

[0120] The figure contains a thin, light gray solid line with sharp fluctuations and a thick, red solid line with gradual changes. The thin, light gray solid line represents the raw Z-axis acceleration acquired by the microelectromechanical system (MEMS) accelerometer. Within the observation interval of 0 to 10 seconds, the signal exhibits large, high-frequency oscillations around the baseline value, caused by road surface bumps. The thick red solid line represents the low-frequency gravity baseline extracted after the raw signal has been processed by a digital low-pass filter.

[0121] As can be seen from the waveform transformation trend, within the time interval of 3 to 4 seconds, the thick red solid line smoothly decreases from a value of 1 to a value of 0.9 and then remains stable. This smooth step-down characteristic objectively reflects the physical process in which the occupant's body mass causes micro-elastic deformation of the local structure of the seat when sitting down or changing posture, which in turn causes a quantifiable shift in the gravity vector projection.

[0122] At this stage, the thick red solid line filters out the high-frequency dynamic alternating components in the light gray waveform, forming a continuous trajectory that only represents the stable gravity state. This chart intuitively reflects that the system can effectively extract the extremely low-frequency gravity baseline for calculating the equivalent static load of occupants from complex broadband road surface excitation noise through filtering, providing solid data support for the dynamic mass coupling correction of the anti-accidental touch algorithm.

[0123] It is important to note that after extracting this baseline, the system needs to convert it from a gravitational acceleration dimension to a mass mechanical dimension. Since the baseline Z-axis offset output when the seat is unloaded is automatically captured and stored during system offline calibration or each initial power-on after vehicle unlocking, when an occupant sits down, the occupant's gravity causes compression of the seat foam and micro-elastic deformation of the bottom metal support, resulting in a quantifiable shift in the extremely low-frequency gravitational acceleration baseline in the Z-axis direction relative to the unloaded reference. The system calculates this differential offset of the baseline in real time and inputs it into a pre-calibrated static transformation matrix to deduce the current occupant's equivalent static load. The current occupant equivalent static load is defined here. It does not refer to the precise biological weight of the occupant in an absolutely static state, but rather to the equivalent static mechanical load that the occupant's body mass, after being transmitted through the seat cushion and skeletal structure, ultimately acts on the local support structure where the target microswitch is located. This parameter directly reflects the interference intensity of the occupant's mass on the local vibration mode.

[0124] Specifically, after completing the load calculation, the low-frequency projection values ​​of the X-axis and Y-axis components of the triaxial spatial acceleration signal are extracted, the absolute ratio of the two is calculated, and the absolute ratio is mapped to the occupant seating posture distribution coefficient. The steps.

[0125] It is also important to note that the occupant's posture during vehicle movement is not a static, unchanging model. For example, during long-distance driving, occupants may lean to one side due to fatigue, or their center of gravity may shift longitudinally forward / backward or laterally when the vehicle accelerates, decelerates, or curves. These real-time changes in posture cause a dynamic shift in the distribution of the occupant's total mass across the various fixed support nodes of the seat frame. If the system only uses the total equivalent static load obtained in the preceding steps for simple mass superposition correction without considering the geometric topological relationship of the mass distribution, then the local coupling mass estimation for microswitches installed on the side panels or lower front edge of the seat will still have a significant deviation.

[0126] For example, to accurately perceive the spatial shift in the occupant's center of gravity distribution, the system utilizes the X-axis component (corresponding to the vehicle's longitudinal coordinate axis) and Y-axis component (corresponding to the vehicle's lateral coordinate axis) output by the same microelectromechanical sensor in the hardware acquisition architecture. It employs the same low-pass filtering strategy as the Z-axis component to filter out dynamic high-frequency components and extract the low-frequency projection values ​​of the X-axis and Y-axis components, representing the static tilt angle. When the occupant's center of gravity deviates from the seat's geometric center, it causes a slight spatial asymmetric deformation in the seat, resulting in a minor change in the spatial attitude angle of the sensor's mounting plane. The gravity vector will then produce corresponding changes in its low-frequency projection components along the X and Y axes.

[0127] Optionally, after extracting the low-frequency projection value, the system calculates the absolute ratio of the two values ​​and inputs it into a specific numerical mapping function to generate a dimensionless occupant seating posture distribution coefficient. The specific calculation formula is as follows:

[0128] ;

[0129] In the formula, This represents the occupant seating posture distribution coefficient. This represents the scaling factor pre-stored in non-volatile memory. This represents the low-frequency projection value extracted from the X-axis component of the triaxial spatial acceleration signal after digital low-pass filtering. This represents the low-frequency projection value extracted from the Y-axis component of the triaxial spatial acceleration signal after digital low-pass filtering. This represents a very small positive number set to prevent overflow errors where the denominator is zero when the microprocessor performs division operations, such as 0.0001.

[0130] It is important to note that the above formula is constructed based on the following logic: a distribution parameter reflecting the spatial center of gravity shift vector is established by utilizing the ratio of the absolute values ​​of lateral and longitudinal low-frequency accelerations. When the occupant's posture involves significant forward, backward, or lateral tilting, the relative strength of the projection components of gravity on the X or Y axis changes, causing this absolute ratio to monotonically increase or decrease. The scaling factor is selected based on the three-dimensional mounting coordinates of the target microswitch on the seat frame, used to converge and limit the calculated physical ratio within an effective weight range that conforms to the local mechanical transmission characteristics of the microswitch. Therefore, this occupant posture distribution coefficient... The influence weight of the occupant's current transient sitting posture on the additional mass distribution of the node where the target microswitch is located was quantitatively and in real time characterized.

[0131] Specifically, after obtaining the two key parameters mentioned above, the equivalent static load based on the current occupants is executed. and occupant seating posture distribution coefficient The temperature-corrected natural resonant frequency is corrected in real time using a preset dynamic coupling model to generate the mass-coupled transient resonant frequency. The steps.

[0132] For example, this pre-defined dynamic coupling model converts macroscopic mass parameters into frequency domain offsets by fusing structural dynamic equations, and the specific calculation formula is as follows:

[0133] ;

[0134] In the formula, This represents the transient resonant frequency of mass coupling after dynamic correction for both temperature and mass. This represents the corrected inherent resonant frequency as described in Example 1. This represents the equivalent basic mass of the seat frame structure as specified by the manufacturer, and all parameters satisfy the dimensionless constraint to ensure that the calculation result inside the square root is a dimensionless attenuation ratio.

[0135] It is important to note that in classical mechanics, the vibration frequency is closely related to the proportional relationship between the system's stiffness and mass. Inside the radical of this formula, the denominator polynomial... Precisely characterizes the current At any given moment, the total equivalent dynamic modal mass after the physical coupling between the occupant mass and the seat frame is considered. This includes the occupant posture distribution coefficient. As an adjustment weight, it is equivalent to the current occupant static load. Perform a product operation to calculate the actual force exerted by the occupant in the current sitting position on the [number of]th [unit]. Effective added mass in the local structure of a micro switch.

[0136] For example, when the occupant's center of gravity moves significantly away from the physical area where the target switch is located, the seating posture distribution coefficient decreases due to the reduced leverage effect. The value of the additional mass contribution decreases accordingly, leading to a lower calculated additional mass contribution value for that region and a relatively smaller total equivalent dynamic modal mass; conversely, if the occupant's center of gravity presses towards the switching region, the additional mass contribution value will significantly increase. The ratio of the equivalent basic mass in the numerator to the total equivalent dynamic modal mass in the denominator, after square root calculation, outputs a dynamic attenuation coefficient less than or equal to 1. The system directly multiplies this attenuation coefficient by the factory-calibrated natural resonant frequency of the microswitch. Above. Based on this physical mathematical model, as the effective equivalent load increases, the corrected mass-coupled transient resonant frequency of the system output increases. It will strictly follow objective physical laws and automatically and smoothly and continuously map to lower frequency bands.

[0137] like Figure 6 The coupled response surface features of occupant load, posture coefficient, and transient resonant frequency are illustrated. In the figure, the horizontal axis represents the equivalent static load in kilograms; the vertical axis represents the posture distribution coefficient in dimensionless units; and the vertical axis represents the calculated transient resonant frequency in Hertz.

[0138] The three-dimensional surface and color distribution in the figure objectively reflect the modulation effect of multivariate physical coupling on the system's resonant frequency. The color of the surface smoothly transitions from deep red in the high-frequency region to deep blue in the low-frequency region.

[0139] Combining the surface geometry and the data in the figure, it can be seen that when the equivalent static load on the horizontal axis approaches 0 kg, the surface is at the highest platform in deep red. At this time, the transient resonant frequency is maintained at about 600 Hz, which corresponds to the basic mechanical characteristics of the seat when it is unloaded.

[0140] As the equivalent static load increases and the sitting posture distribution coefficient increases simultaneously, the surface representing the frequency value exhibits a significant nonlinear downward trend. Specifically, when the equivalent static load reaches 100 kg and the sitting posture distribution coefficient reaches its maximum of 1.0, the surface dips significantly into the dark blue region, and the transient resonant frequency drops to around 245 Hz.

[0141] The steep descent of this spatial surface intuitively and quantitatively reveals the inherent frequency shift caused by the direct intervention of the seat vibration system by the body mass of a heavily loaded occupant as an additional mass block when the occupant sits down and the center of gravity of the body is significantly pressed against the target micro-switch area.

[0142] The chart provides ample data support, demonstrating that this technical solution can perceive changes in the boundaries of complex mechanical environments in real time through underlying data-driven approaches, and smoothly adjust the frequency band matching search window accordingly. This avoids the missed detection phenomenon caused by the use of fixed frequency parameters in traditional anti-accidental touch systems, and enhances the reliability of the system's state identification under complex loads and variable sitting postures.

[0143] Specifically, after calculating the aforementioned transient frequency, the system performs the mass-coupled transient resonant frequency calculation. The updated target matching center frequency is used to replace the pre-stored mechanical resonance fingerprint in the step of generating resonance flags, so as to dynamically follow the shift of the real resonance frequency caused by changes in occupant mass.

[0144] For example, in the short-time Fourier transform spectrum analysis and matching module described in Embodiment 1, the microcontroller originally relied on the static reference frequency in the pre-stored mechanical resonance fingerprint to construct a fixed frequency band search window to capture the real-time peak frequency. However, in this embodiment, the microcontroller utilizes the computing power of the central processing unit to continuously iterate the aforementioned dynamic coupling model at a preset update cycle, such as a background timer task every 100 milliseconds, continuously calculating the mass-coupled transient resonant frequency deeply bound to the current occupant state. .

[0145] Optionally, in the preceding moments of each frequency domain feature matching, the microcontroller's memory management unit directly overwrites and replaces the original fingerprint base frequency parameter in the global static variables with the newly calculated mass-coupled transient resonant frequency. This means that the frequency band matching window of the short-time Fourier transform algorithm will dynamically drift with this new mass-coupled transient resonant frequency as the target. When the vehicle is traveling on a rough and bumpy road, the actual road vibration force induces the true resonant frequency of the seat frame under the current occupant's heavy load. Since this true mechanical frequency also decreases with the intervention of the occupant's physical mass, the matching search window dynamically drifted by the algorithm can accurately capture this reduced physical frequency peak. The system then calculates the resonance matching degree of the real-time peak frequency based on the updated parameters, thereby ensuring that even under extreme heavy load, light load switching, or frequent alternation of complex sitting postures, the anti-accidental touch system can still accurately identify the signal and output a true and valid resonance flag.

[0146] In terms of existing technologies, traditional anti-mistouch suppression algorithms and current compensation techniques generally suffer from a structural flaw: over-reliance on offline calibration data, treating the mechanical characteristic parameters of the seat system as time-invariant absolute constants. This isolated, static processing logic leads to a disconnect between the fixed parameter model and the real physical world when physical boundary conditions of the application environment change, such as changes in occupant weight or seating posture, resulting in serious false alarms in the anti-mistouch system. The technical solution described in this embodiment, however, deeply mines the information contained in the low-frequency DC response of existing three-axis spatial accelerometers, constructing an optimization and correction mechanism based on the dual-coupling physical equations of occupant mass and seating posture, without increasing vehicle manufacturing costs or hardware sensor complexity.

[0147] The overall technical solution achieves the following effect: the system realizes highly sensitive real-time perception and adaptive dynamic reconstruction of changes in the boundary of its complex mechanical environment. Whether the seat is in an unloaded standby state, carrying children or adults with significant weight differences, or the occupant frequently changes their leaning posture during long-distance travel, causing a shift in the center of gravity, this dynamic fingerprint correction step, driven by continuous data at the underlying level, can adjust the core judgment benchmark of the anti-mistouch algorithm in real-time and smoothly to an optimal level that matches the current macroscopic physical reality. This effectively avoids the failure of the anti-mistouch mechanism caused by a significant deviation in the resonant frequency due to changes in the system's equivalent modal mass, significantly reduces the probability of seat malfunctions under complex operating conditions, greatly expands the engineering applicability of the compensation algorithm in different occupant characteristic groups, and improves the long-term operational safety, reliability, and robustness of state judgment of the system.

[0148] Example 3:

[0149] Specifically, after matching the real-time spectral features with the pre-stored mechanical resonance fingerprint to generate a resonance flag, the method in this embodiment also includes a step to suppress pseudo-resonance caused by global periodic excitation.

[0150] It is important to note that in actual vehicle operating environments, the causes of high-amplitude vibrations at specific frequencies at the physical nodes where microswitches are located have a dual nature in terms of physics and dynamics. The first reason is that the local metal springs or cantilever beam structure of the microswitch undergoes independent mechanical self-excited resonance, i.e., local true resonance. The mode shape of this resonance is limited to the inside of the switch and its adjacent small areas, which is the core physical mechanism that causes high-frequency bouncing of the contacts and accidental contact.

[0151] The second reason is that the external excitation source applies a strong, constant-frequency periodic excitation force to the entire vehicle chassis and seat frame, causing the entire seat frame, as a macroscopic approximate rigid body, to undergo overall translational or torsional vibration, i.e., global forced vibration. In single-node time-frequency domain analysis, both of these physical phenomena will exhibit the same frequency energy peak on the amplitude spectrum of the short-time Fourier transform, making traditional single-point spectrum matching algorithms prone to serious state misjudgments.

[0152] For example, when a vehicle travels on an unpaved road surface with continuous, equally spaced undulations, such as a rural washboard road or a matrix of continuous speed bumps, or when the vehicle's engine or driveshaft operates within a specific high-speed range, causing resonance, and the fundamental frequency of this external forced excitation force happens to fall within the inherent resonant frequency band matching window of the microswitch, the single-point accelerometer of the target microswitch will also output a high-amplitude corresponding frequency component. Based on the single-point amplitude spectrum, the system will determine that both the frequency and amplitude meet the standards, thus incorrectly classifying it as local self-excited resonance (this phenomenon is defined in engineering as global pseudo-resonance), and consequently incorrectly triggering the resonance-specific dynamic compensation model.

[0153] To address the spatial aliasing misjudgment problem caused by the lack of single-point observation dimensions, this embodiment introduces a cross-node spatial topology and phase consistency identification mechanism based on a multi-source distributed acquisition architecture.

[0154] Specifically, the step for suppressing pseudo-resonance caused by global periodic excitation includes: when the resonance flag of the i-th microswitch is determined to be in an effective state, acquiring the triaxial spatial acceleration signal corresponding to the m-th microswitch that is farthest from the i-th microswitch in spatial straight line distance as a global reference signal.

[0155] Optionally, during the acquisition of the global reference signal, the system's central processing unit (CPU) calls the three-dimensional spatial coordinate matrix of the seat frame pre-stored in read-only memory (ROM). This matrix records the local coordinates of all distributed microelectromechanical system (MEMS) accelerometers relative to the geometric center of the seat frame. Using the coordinates of the i-th microswitch that triggered the resonance flag as the spatial center, the system iterates through and calculates the spatial Euclidean distances to all other untriggered and triggered microswitches, and selects the node with the largest distance value, calibrating it as the m-th microswitch.

[0156] It is important to note that choosing the physical node with the greatest linear distance in space as the global reference signal source has a rigorous structural dynamics basis. In continuum mechanics, the excitation energy of local self-excited resonance, when transmitted outward, is subject to internal damping of the seat frame's metal material, abrupt changes in geometric cross-section, and severe dissipation by connecting hinges, resulting in an extremely short spatial coherence length; the greater the distance, the weaker the associated vibration response induced by local resonance. Conversely, in the case of global periodic forced excitation at the vehicle level, the energy wavelength of low-frequency or specific frequency bands is much larger than the physical dimensions of the seat frame. The entire frame is in a vibration field with the same phase or a fixed linear phase difference, and the two farthest nodes will still maintain a very high kinematic correlation. Therefore, the geometric constraint of "greatest linear distance in space" maximizes the observational differences between local resonance and global vibration in spatial phase topology.

[0157] Specifically, after anchoring the global reference signal source, the step of performing Hilbert transform on the three-axis spatial acceleration signal of the i-th micro switch and the global reference signal respectively, and extracting the instantaneous phase of the two at the current moment.

[0158] It is also important to note that in traditional digital signal processing, the conventional techniques for obtaining phase information are Discrete Fourier Transform (DFT) or Fast Fourier Transform (FFT). However, the core idea of ​​the Fourier Transform is to perform a static global orthogonal basis projection on the signal segment within the entire analysis time window. Its output phase spectrum characterizes the average phase delay of that frequency point within the entire time window, lacking sufficient time resolution to depict the transient phase evolution trajectory under non-stationary vibration environments. To accurately capture the dynamic phase difference between two spatial nodes within the same extremely short time slice, this embodiment abandons the Fourier phase spectrum and instead employs the Hilbert Transform mechanism, which has continuous analytical capability in the time domain.

[0159] For example, the Hilbert transform constructs a physically meaningful analytic signal in the complex plane by taking the original real-domain acceleration signal as its real part and performing a 90-degree phase shift on all its frequency components to obtain its imaginary part. The underlying mathematical model for constructing this analytic signal and extracting its instantaneous phase is calculated using the following formula:

[0160] ;

[0161] in, For the first The complex domain analytic signal corresponding to each micro switch For the first The time-domain synthesized acceleration signal corresponding to each microswitch after removing the DC bias. For Hilbert's transformation mathematical operators, To obtain the instantaneous amplitude envelope of the solved analytic signal, To solve the first The instantaneous phase of the triaxial spatial acceleration signal of a microswitch in the continuous time domain. Through the same mathematical operation logic, the system synchronously obtains the instantaneous phase corresponding to the global reference signal, thereby generating two high-resolution phase change trajectories that are strictly aligned on the time axis.

[0162] Specifically, after successfully extracting the instantaneous phase sequences of two spatial nodes, the calculation of the first phase sequence is performed within a preset sliding time window. The micro switch and the first Spatial phase lock value between microswitches The steps are as follows. It's also important to note that the spatial phase-locking value is a quantitative indicator for complex systems derived from nonlinear dynamics and EEG network topology analysis. It's used to rigorously evaluate the degree of phase synchronization or phase locking between two dynamic sequences with time-evolutionary characteristics within a specific frequency band. Unlike traditional Pearson correlation coefficients or amplitude-based cross-correlation functions, the spatial phase-locking value completely eliminates the interference of instantaneous amplitude envelopes on the results, statistically averaging the direction vector of the instantaneous phase difference only on a unit complex circle. Therefore, it is highly immune to amplitude attenuation caused by differences in sensor mounting stiffness, and can reveal the purest kinematic interference characteristics.

[0163] For example, the system's central processing unit acquires a data sequence in the discrete time domain according to a preset sampling frequency, and the spatial phase-locked loop value... The calculation formula is:

[0164] ;

[0165] in, This is the spatial phase-locked value. The total number of discrete sampling points within the sliding time window. The dedicated imaginary unit for complex plane analytic signals. For the first A micro switch in The instantaneous phase of a moment, For the first A micro switch in The instantaneous phase of a moment.

[0166] Optionally, the total number of discrete sampling points in the sliding time window in the above formula... The selection is not arbitrary; rather, a mapping relationship must be established with the period of the target's center frequency. This is to ensure sufficient mathematical confidence in the statistical results while maintaining the response speed to transient shocks. The corresponding time span typically covers three to five complete vibration cycles of the target frequency.

[0167] During the execution of this formula, the system first calculates each discrete time step. The instantaneous phase difference between the two Subsequently, this instantaneous phase difference is substituted into Euler's formula as the independent variable to generate a direction vector with a constant magnitude of one in the complex plane. Next, the system processes the data within the sliding time window. Perform vector summation on each direction vector, take the absolute value of the complex modulus of the summation result, and finally divide by the total number. Perform a normalization mapping. After the rigorous mathematical mapping described above, The calculation results are strictly defined in the interval [0,1].

[0168] like Figure 7 The figure illustrates the coherent scatter distribution characteristics of spatial phase-locked loop values ​​on the unit circle of the complex plane. The horizontal axis represents the real part of the complex plane analytic signal, in dimensionless units; the vertical axis represents the imaginary part of the complex plane analytic signal, also in dimensionless units.

[0169] The figure includes a black unit circle boundary, as well as a scatter plot representing two different physical conditions and a composite vector arrow. The blue scatter plots represent local resonant phase differences, which are randomly and uniformly distributed on the circumference of the unit circle. This reflects that within a preset sliding time window under the local real resonant condition, there is no stable time synchronization relationship between the instantaneous phase difference between the target microswitch that triggers the resonant flag and the global reference microswitch that is spatially furthest away.

[0170] Since the direction vectors represented by the blue scatter dots are distributed at various angles, significant mutual cancellation occurs during vector summation in the complex plane, resulting in the extremely short blue arrow in the figure, i.e., the local resonance synthesis vector. The length of this vector indicates that the calculated spatial phase-locked value at this time approaches 0.1, which is less than the preset spatial coherence threshold of 0.7 to 0.85. Based on this, the system determines it as a local true resonance and maintains the effective state of the corresponding switching resonance flag to perform dynamic compensation.

[0171] The red dots represent the global vibration phase difference, which exhibits a highly concentrated clustered distribution in the first quadrant of the complex plane. This reflects that under the condition of global periodic forced vibration, the entire seat frame is driven by the external road surface excitation force to produce macroscopic translation, resulting in the two farthest micro-switch nodes still maintaining a constant phase delay.

[0172] Because the red scatter points are highly aligned, the vector summation process produces a strong superposition in the same direction, generating the red arrow in the figure that extends significantly and approaches the boundary of the unit circle, i.e., the global vibration synthesis vector. The length of this vector indicates that the spatial phase-locked value has reached a high value range of 0.9 or above, which is greater than the preset spatial coherence threshold.

[0173] Based on this high spatial phase-locked loop value, the system determines that it is currently in a global forced vibration condition, and then forcibly resets the resonance flag of the corresponding microswitch to an invalid state, thus suppressing false resonance misjudgment and incorrect model calling caused by periodic road bumps.

[0174] Specifically, after completing the high-speed parallel mathematical calculation of the spatial phase-locked value, the system switches to a logic decision branch to execute the calculation of the spatial phase-locked value. The step involves comparing the data with a preset spatial coherence threshold. This spatial coherence threshold is determined by the system calibration engineer through cluster analysis algorithms during full-road bench durability testing before the vehicle rolls off the production line, and is typically set within the strong correlation range of 0.7 to 0.85.

[0175] For example, if the spatial phase-locked value If the spatial coherence threshold is greater than or equal to the threshold, it is determined that the current condition is a global forced vibration condition, and the first... The resonant flag of a microswitch is forcibly reset to an invalid state.

[0176] It is also important to note that when the calculated spatial phase-locked value approaches 1, its underlying mathematical meaning is: within the entire sliding time window, the first... The microswitch is located at the furthest point. The instantaneous phase difference between the microswitches remains constant (even though their absolute phases are changing drastically). In classical mechanical vibration theory, the only reasonable physical explanation for the long-term phase lock between two spatially distant nodes is that the seat frame, as a rigid structure, is undergoing macroscopic low-order forced vibration driven by a strong external road surface excitation source. At this point, even if the first... The frequency collected by the single-point sensor of the microswitch happens to fall within the matching window, and this is not a local resonance of the switch reed cantilever beam itself. Therefore, the system should identify this "pseudo-resonance" interference and forcibly reset the resonance flag to invalid (logic 0) through a software interrupt, suppressing the erroneous triggering of the resonance-specific dynamic compensation model, ensuring that the compensation system continues to call the non-resonance scenario-specific cross-domain transfer function model to perform conventional filtering, and preventing overcompensation from masking the actual pressing operation.

[0177] Optionally, if the spatial phase-locked value If the spatial coherence threshold is less than the threshold value, it is determined that the current state is a local true resonance condition, and the first condition is maintained. The resonant flag of each microswitch is in an active state.

[0178] For example, when the spatial phase-locked value is small, such as approaching 0 or being in the low to medium range, it indicates that within the sliding time window, the calculated value... The directional vectors exhibit a scattered, disordered, yet uniformly distributed state on the unit circle of the complex plane, resulting in significant mutual cancellation during vector summation. This reflects the ... The vibration phase of the microswitch does not exhibit any stable time synchronization with the vibration phases of other distal nodes in the main frame of the seat. At the dynamic level, this demonstrates that the internal mechanical microstructure of the target microswitch is effectively excited by energy of a specific frequency, entering a high-frequency self-excited oscillation independent operating state, and its vibration behavior has deviated from the rigid body motion constraints of the macroscopic frame. Since this is confirmed as genuine local physical resonance, the system maintains the valid state of this resonance flag and continues to authorize the hierarchical dynamic compensation solution module to call the high-intensity resonance-specific dynamic compensation model to strip away the severely distorted interference current component.

[0179] In terms of existing technology, current physical button anti-mistouch systems are limited to isolated point-state signal acquisition methods. This point-to-point threshold comparison or single-channel frequency domain filtering mechanism has an inherent system limitation: it cannot distinguish between local high-frequency self-excited modes and macroscopic road surface periodic forced excitation in a multi-dimensional physical space. Once a vehicle enters a continuously uneven road surface, conventional systems are highly susceptible to widespread algorithm failure due to energy overflow at a single frequency point.

[0180] The technical solution described in this embodiment utilizes an existing multi-node hardware acquisition network to successfully upgrade the original one-dimensional time-frequency domain analysis dimension to a four-dimensional tensor analysis space composed of time, frequency, spatial topology, and instantaneous phase. Through microsecond-level extraction of the instantaneous phase using Hilbert transform, combined with the quantitative evaluation of complex plane coherence using spatial phase-locked loop (PLL) indices, the system endows the control core with the ability to identify both macroscopic rigid body motion of the vehicle and microscopic self-excited resonance of microswitches without adding any external physical structural components. This overall solution effectively overcomes the problem of compensation logic errors caused by "pseudo-resonance," significantly reducing the false positive rate of the anti-misoperation algorithm on rough paved roads or in the high-speed range of the engine, enabling the dynamic compensation system to maintain a high degree of control independence and state reliability even under extremely complex driving conditions.

[0181] Example 4:

[0182] like Figure 8 As shown, this embodiment provides a dynamic compensation system for preventing accidental touches of a vehicle seat micro switch.

[0183] When the vehicle's operating conditions cause the external road surface vibration frequency to approach or coincide with the microswitch's own mechanical natural frequency, the internal metal spring of the microswitch is prone to mechanical resonance, causing the contacts to bounce at high frequency, thus generating an interference current that highly overlaps with the characteristics of a manual press. At the same time, due to the long-term service of the vehicle, the contact resistance of the microswitch contact surface will age and change, causing the traditional fixed threshold anti-accidental touch strategy to fail, triggering the seat adjustment motor to start accidentally.

[0184] To address the aforementioned physical vibration coupling and electrical aging drift issues, this embodiment constructs a hardware and software collaborative underlying control system.

[0185] A dynamic compensation system for preventing accidental activation of a vehicle seat microswitch includes: a synchronous acquisition module, a feature extraction and matching module, a parameter acquisition and scene classification module, a hierarchical compensation calculation module, a joint state determination module, and an output and fault handling module. In practical physical applications, these functional modules are primarily integrated onto the printed circuit board (PCB) of the vehicle seat control unit (SCU), relying on a multi-core microcontroller (MCU) conforming to automotive-grade safety standards (such as ASIL level) and its peripheral hardware circuitry.

[0186] Specifically, this system includes a synchronous acquisition module, which is used to synchronously acquire the three-axis spatial acceleration signal and the real-time loop current signal corresponding to each micro switch through hardware triggering.

[0187] In the actual hardware device, the synchronous acquisition module mainly consists of a distributed microelectromechanical system (MEMS) accelerometer array, a precision current detection circuit, and a hardware timer group of the microcontroller. The MEMS accelerometer is rigidly fixed to the metal frame node of the seat adjacent to the microswitch using surface mount technology, and is used to sense the spatial excitation force in this area in real time. The current detection circuit includes a low temperature coefficient shunt resistor connected in series in the microswitch signal loop, and a differential operational amplifier connected across the shunt resistor.

[0188] To achieve strict time synchronization, the microcontroller is equipped with a general-purpose timer that generates a hardware trigger pulse at set intervals. This pulse simultaneously triggers the analog-to-digital converter (ADC) to sample the voltage output of the differential operational amplifier via the internal bus, and triggers the Serial Peripheral Interface (SPI) bus to read the data register of the MEMS sensor. This purely hardware-based triggering mechanism eliminates the scheduling delay of the software operating system, enabling microsecond-level alignment of the three-axis spatial acceleration signal and the real-time loop current signal on the time axis.

[0189] Furthermore, the system includes a feature extraction and matching module, which is used to extract real-time spectral features from the triaxial spatial acceleration signal, match the real-time spectral features with a pre-stored mechanical resonance fingerprint, and generate a resonance flag bit.

[0190] At the hardware computing layer, this module relies on the integrated digital signal processing (DSP) coprocessor or floating-point unit (FPU) within the microcontroller. The microcontroller transfers the acquired triaxial spatial acceleration data to static random access memory (SRAM) and uses a Fast Fourier Transform (FFT) hardware acceleration engine to perform windowed time-frequency transformation on the signal to extract the current real-time peak frequency. Simultaneously, the system's non-volatile memory (such as EEPROM or Data Flash) pre-stores the mechanical resonance fingerprint (including the fundamental natural frequency and temperature correction curve) calibrated before the microswitch leaves the factory. The microcontroller retrieves this pre-stored fingerprint data, calculates the matching degree between the real-time spectral characteristics and the mechanical resonance fingerprint, and when resonance is detected, sets a specific bit in the internal register to a valid state, i.e., generates a resonance flag.

[0191] Furthermore, this system includes a parameter acquisition and scene classification module, used to acquire the road condition type of the vehicle where the micro switch is located and the service life level of the micro switch, and calculate the online estimated contact resistance corresponding to the real-time current signal of the circuit; combined with the road condition type, the service life level of the micro switch, the online estimated contact resistance and the resonance flag, a four-dimensional scene classification result is generated.

[0192] In practical applications, parameter acquisition relies on the vehicle's communication network. The microcontroller receives road roughness messages broadcast from the Electronic Stability Program (ESP) or active suspension control unit via the Controller Area Network (CAN) or CAN-FD transceiver to obtain the road condition type; and assesses the service life level of the microswitches by reading the odometer mileage or vehicle power-on cumulative duration messages from the vehicle's central gateway node.

[0193] Meanwhile, the microcontroller uses another ADC channel to monitor the power supply voltage of the switching circuit in real time. Combined with the previously acquired steady-state current data, it calculates the current online estimated contact resistance of the microswitch based on Ohm's law. After acquiring the multidimensional physical quantities, the microcontroller maps the current operating condition to a specific scene classification space according to a preset logic decision tree or look-up table algorithm, and outputs a four-dimensional scene classification result.

[0194] Furthermore, the system includes a hierarchical compensation solution module, which performs hierarchical dynamic compensation solution based on the four-dimensional scene classification results to obtain the preliminary state and the accurate pure actual current.

[0195] This module is primarily executed by the microcontroller's main computing core. Different model parameter areas are partitioned in the system's program memory (Flash) for different four-dimensional scene classification results. When the scene indicates a non-resonant condition, the main computing core calls the Taylor impedance polynomial model to calculate the resistance modulation component caused by low-frequency vibration; when the scene indicates a resonant condition, the main computing core calls the dynamic resonance compensation model containing the Sigmoid nonlinear function. Through high-frequency algebraic operations, the microcontroller calculates the current interference current component and subtracts this interference current component from the real-time current signal of the loop in the digital domain, thereby outputting a precise and pure actual current free from mechanical vibration coupling noise.

[0196] Furthermore, the system includes a joint state determination module, which is used to construct the state confidence vector of all microswitches based on the accurate pure actual current, establish a multi-objective optimization objective function in combination with the preset multi-switch state correlation matrix, and solve for the globally optimal switch state combination.

[0197] In practical electronic control units, the precise, pure actual currents of each microswitch are aggregated to form a multi-dimensional state vector. The multi-switch state correlation matrix, stored in read-only memory, is established based on ergonomic constraints, such as the inability of a human hand to simultaneously flip two interfering switches in opposite directions. Within the control cycle of the real-time operating system (RTOS), the microcontroller solves for this objective function using numerical optimization algorithms such as sequential quadratic programming or gradient descent. Due to the large number of matrix multiplication and addition operations involved, this step is typically performed in a separate algorithm execution thread within the microcontroller to avoid blocking the underlying signal acquisition. The output of the solution is the globally optimal combination of switch states that conforms to both physical constraints and electrical logic.

[0198] Finally, the system includes an output and fault handling module, which outputs the globally optimal switch state combination to the vehicle controller and performs a graded fault handling process in conjunction with the three-axis spatial acceleration signal, the real-time loop current signal and the resonance flag.

[0199] At the physical execution layer, this module connects to the power drive of the seat adjustment motor. If the microswitch status is used for local control, the microcontroller outputs a pulse width modulation (PWM) signal to the half-bridge or full-bridge motor driver chip to drive the seat's slide rail, backrest, or height adjustment motors. If the status needs to be used globally, the status message is sent to the vehicle control unit (VCU) via a CAN transceiver. Simultaneously, the fault handling module has hardware-level intervention capabilities. For example, when a level three fault (physical contact adhesion) is determined based on acceleration and current signals, the microcontroller will directly disconnect the enable pin of the motor driver chip, or cut off the power supply bus for the entire seat motor via a high-side switch. For minor aging or poor contact, the system records a diagnostic fault code (DTC) in the EEPROM, which can be read by after-sales repair equipment via the Unified Diagnostic Service (UDS) protocol.

[0200] In summary, this vehicle seat microswitch anti-misoperation dynamic compensation system, through precise synchronous data acquisition from multiple hardware sensors and combined with multi-model hierarchical compensation calculations within the microcontroller for four-dimensional scenarios, establishes an electromechanical coupling identification device independent of traditional simple electrical filtering in practical applications. When the vehicle encounters severe road conditions and vibrations, this system can effectively separate the glitch noise caused by mechanical resonance from the passenger's actual pressing intention. Even in the later stages of the component's lifespan, it can overcome parameter drift caused by contact aging through online resistance estimation and scene adaptive classification. This system improves the control reliability of vehicle seat adjustment functions in complex physical environments, reduces the probability of abnormal motor start-up, and possesses significant engineering application value.

[0201] Example 5:

[0202] Corresponding to the above embodiments, the present invention also proposes an electronic device.

[0203] like Figure 9 The diagram shows a structural schematic of an electronic device according to the present invention. The electronic device 100 includes a processor 101 and a memory 103. The processor 101 and the memory 103 are connected, for example, via a bus 102. Optionally, the electronic device 100 may further include a transceiver 104. It should be noted that in practical applications, the transceiver 104 is not limited to one unit, and the structure of this electronic device 100 does not constitute a limitation on the embodiments of the present invention.

[0204] Processor 101 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. Processor 101 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0205] Bus 102 may include a pathway for transmitting information between the aforementioned components. Bus 102 may be a PCI bus or an EISA bus, etc. Bus 102 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 9 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0206] The memory 103 stores a computer program corresponding to a method for preventing accidental touches of a vehicle seat microswitch according to the above embodiments of the present invention. This computer program is controlled and executed by the processor 101. The processor 101 executes the computer program stored in the memory 103 to implement the content shown in the aforementioned method embodiments.

[0207] Among them, electronic devices 100 include, but are not limited to: mobile terminals such as laptops and PADs (tablet computers) and fixed terminals such as desktop computers. Figure 9 The electronic device 100 shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0208] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A dynamic compensation method for preventing accidental touches of a micro switch for a vehicle seat, characterized in that, include: The three-axis spatial acceleration signal and the real-time loop current signal corresponding to each micro switch are synchronously acquired through hardware triggering. The real-time spectral features are obtained by extracting features from the triaxial spatial acceleration signal. The real-time spectral features are then matched with the pre-stored mechanical resonance fingerprint to generate a resonance flag. The road condition type of the vehicle where the microswitch is located and the service life level of the microswitch are obtained. The power supply voltage and inherent resistance of the microswitch's switching circuit are also obtained. Based on the power supply voltage of the switching circuit, the inherent resistance of the circuit, and the steady-state average value of the accurate and pure actual current when the microswitch is closed, the online estimated contact resistance corresponding to the real-time current signal of the circuit is calculated. The calculation formula is as follows: ； in, For the online estimated contact resistance, Provide the power supply voltage for the switching circuit. This represents the steady-state average value of the precise, pure, actual current when the microswitch is closed. The inherent resistance of the line; By combining the road condition type, the service life level of the micro switch, the online estimated contact resistance, and the resonance flag, a four-dimensional scene classification result is generated. Based on the four-dimensional scene classification results, hierarchical dynamic compensation calculations are performed to obtain the preliminary state and the precise pure actual current, specifically including: When the resonance flag is invalid, the compensation impedance is calculated by calling the non-resonant scenario-specific cross-domain transfer function model: ； in, To compensate for impedance, The transient energy envelope is extracted based on the triaxial spatial acceleration signal. , , These are nonlinear coefficients. This is the contact resistance correction factor; based on the compensation impedance. The non-resonant interference current component is calculated from the power supply voltage of the switching circuit, and the non-resonant interference current component is subtracted from the real-time current signal of the circuit to obtain the accurate and pure actual current. When the resonance flag is active, obtain the insulation resistance of the microswitch in the open state; call the resonance-specific dynamic compensation model to calculate the resonance interference current component: ； in, This is the resonant interference current component. The insulation resistance of the microswitch in the off state. A nonlinear coefficient characterizing the abrupt change rate of contact separation impedance. This represents the contact threshold displacement. The relative displacement of the contact at resonance is calculated based on the corrected resonant gain coefficient and the corrected resonant damping ratio; the resonant interference current component is subtracted from the real-time current signal of the circuit to obtain the accurate and pure actual current; Based on the precise and pure actual current, construct the state confidence vector of all microswitches, and combine it with the preset multi-switch state correlation matrix to establish a multi-objective optimization objective function, and solve to obtain the global optimal switch state combination. The globally optimal switch state combination is output to the vehicle controller, and a graded fault handling flow is executed in conjunction with the three-axis spatial acceleration signal, the real-time loop current signal and the resonance flag.

2. The method according to claim 1, characterized in that, The step of matching the real-time spectral features with a pre-stored mechanical resonance fingerprint to generate a resonance flag bit includes: The three-axis spatial acceleration signals are vector synthesized to obtain a synthesized acceleration signal; Perform a sliding window short-time Fourier transform on the synthesized acceleration signal to extract the real-time peak frequency and peak amplitude; Obtain the current ambient temperature and calculate the corrected natural resonant frequency based on the current ambient temperature; Calculate the resonance matching degree between the real-time peak frequency and the corrected inherent resonant frequency; When the resonance matching degree is greater than or equal to the first matching threshold and the peak amplitude is greater than or equal to the first amplitude threshold, the corresponding micro switch is determined to have entered the resonance condition, and the resonance flag position is set to the valid state; otherwise, the resonance flag position is set to the invalid state.

3. The method according to claim 2, characterized in that, The formula for calculating the corrected natural resonant frequency is as follows: ； in, This is the corrected natural resonant frequency. This is the inherent resonant frequency calibrated at the factory for the microswitch. and These are the first-order and second-order temperature correction factors, respectively. The current ambient temperature, This is the preset reference ambient temperature.

4. The method according to claim 2, characterized in that, The mechanical resonance fingerprint includes an aging correction coefficient, and the method further includes a step of self-calibrating the aging correction coefficient: Sample data under resonance conditions are collected at a preset period, and the aging correction coefficient is updated using the recursive least squares method. The calculation formula is as follows: ； in, This is the updated aging correction factor. This is the current aging correction factor. For Kalman gain, For the first The real-time peak frequency extracted when the system first enters the resonance condition. This is the inherent resonant frequency calibrated at the factory for the microswitch. This corresponds to the current service life of the micro switch.

5. The method according to claim 1, characterized in that, The expression for the multi-objective optimization objective function is: ； in, The objective function value, For the first The weight of each microswitch The confidence level is the value in the state confidence vector. For the current moment, the first A state mapping value for a microswitch, wherein the state mapping value includes two discrete states: open and closed. For the current moment The state mapping value of a micro switch The elements of the preset multi-switch state correlation matrix, The consistency penalty coefficient, This is the continuous penalty coefficient. The penalty coefficient for the resonant state. For the previous moment The state mapping value of a micro switch. This corresponds to the resonant flag mapping value of the microswitch.

6. The method according to claim 1, characterized in that, Before constructing the state confidence vector of all microswitches based on the accurate pure actual current, the method further includes: When the resonant flag is in an active state, the threshold for the state determination duration of the corresponding micro switch is updated from a first duration to a second duration, wherein the second duration is longer than the first duration. Calculate the total harmonic distortion of the precise pure actual current, and determine whether the signal is a real trigger signal or a resonance interference signal based on the distribution of the total harmonic distortion within a preset interval defined by a first distortion threshold and a second distortion threshold, wherein the second distortion threshold is greater than the first distortion threshold. The resonant flags of three adjacent microswitches in the detection space are checked. If the resonant flags of the adjacent microswitches are all invalid, the confidence level of the current microswitch state is confirmed by combining the judgment result of the total harmonic distortion.

7. A dynamic compensation system for preventing accidental touches of a micro switch for a vehicle seat, characterized in that, include: The synchronous acquisition module is used to synchronously acquire the three-axis spatial acceleration signal and the real-time current signal of the loop corresponding to each micro switch through hardware triggering. The feature extraction and matching module is used to extract features from the triaxial spatial acceleration signal to obtain real-time spectral features, and match the real-time spectral features with the pre-stored mechanical resonance fingerprint to generate a resonance flag bit. The parameter acquisition and scene classification module is used to acquire the road condition type of the vehicle where the micro switch is located and the service life level of the micro switch, and calculate the online estimated contact resistance corresponding to the real-time current signal of the circuit; and combine the road condition type, the service life level of the micro switch, the online estimated contact resistance and the resonance flag to generate a four-dimensional scene classification result. The hierarchical compensation solution module is used to perform hierarchical dynamic compensation solution based on the four-dimensional scene classification result to obtain the preliminary state and the accurate pure actual current. Specifically, it is used to: when the resonance flag bit in the four-dimensional scene classification result is invalid, call the non-resonant scene dedicated cross-domain transfer function model to calculate the compensation impedance, calculate the non-resonant interference current component based on the compensation impedance and the switch circuit power supply voltage, and subtract the non-resonant interference current component from the real-time current signal of the circuit to obtain the accurate pure actual current. When the resonance flag in the four-dimensional scene classification result is in an effective state, the insulation resistance of the micro switch in the open state is obtained, the resonance-specific dynamic compensation model is called to calculate the resonance interference current component, and the resonance interference current component is subtracted from the real-time current signal of the circuit to obtain the accurate and pure actual current. The joint state determination module is used to construct the state confidence vector of all microswitches based on the accurate pure actual current, and to establish a multi-objective optimization objective function by combining the preset multi-switch state correlation matrix, and to solve for the global optimal switch state combination. The output and fault handling module is used to output the globally optimal switch state combination to the vehicle controller, and to perform a graded fault handling process in conjunction with the three-axis spatial acceleration signal, the real-time loop current signal and the resonance flag.

Images