A vehicle dynamic weighing and load estimation method

Abstract

The present application relates to the technical field of load estimation, and discloses a vehicle dynamic weighing and load estimation method, which comprises the following steps: firstly, collecting the vehicle speed, suspension tower top displacement and road space pitch signals by a signal acquisition unit; calculating the main excitation frequency and angular frequency according to the pitch and vehicle speed by a road excitation processing unit; extracting the compliance frequency response function of the suspension response by a modal and parameter unit, and obtaining the characteristic frequency, bandwidth and inverse frequency response function through parameterization, and meanwhile, constructing a frequency weight function; generating the mode shape ratio and combining the suspension displacement to form a synthetic observation signal according to the modal result by a synthetic observation unit; calculating the load frequency domain estimation by the inverse frequency response function, main excitation angular frequency, frequency weight and static parameters by a load estimation unit; and obtaining the net load signal by signal conversion, filtering and constraint processing of the load frequency domain estimation by a net load unit.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of load estimation, more particularly, it relates to a vehicle dynamic weighing and load estimation method. BACKGROUND

[0002] With the development of new energy vehicle technology, the structural integration design of battery pack (i.e. the battery pack directly as the load-bearing structure of the vehicle body floor) is widely adopted. This structural design significantly improves the overall stiffness and load-carrying capacity of the vehicle body, but at the same time, it also changes the bending and torsional response characteristics of the vehicle. When the vehicle is driving on a road with obvious geometric repeating units, the excitation of the suspension system by the road presents a narrow-band periodic characteristic, and its main excitation frequency is determined by the vehicle speed and the periodic length of the road. The interaction between this excitation and the new vehicle body dynamics characteristics brought by the structural battery floor forms a completely different response mode from the traditional steel vehicle body.

[0003] In actual operation, the vehicle often passes through the above-mentioned endurance road section at a speed of 20km / h to 30km / h, at which time the excitation frequency falls into the main frequency band of the vehicle body torsion and bending response. Due to the uneven distribution of materials and the difference in stiffness in all directions of the structural battery floor, the local deformation of the vehicle body in this frequency range will change abnormally, causing abnormal peaks and valleys in the dynamic flexibility at the suspension support points. This change in local dynamic response directly affects the suspension stroke signal, airbag pressure signal and other measured quantities used for load estimation, resulting in a significant increase in the deviation between the estimated result and the actual load, specifically:

[0004] The traditional load estimation method is realized based on the linear relationship between the suspension displacement and the wheel load, and the mathematical mapping is established through static test or low-frequency calibration. However, in the vehicle with structural battery floor, the dynamic flexibility of the vehicle body structure changes significantly with frequency, and the local stiffness suddenly rises or falls at a certain excitation frequency, so that the effective displacement of the suspension support point no longer reflects the change of the wheel load linearly. Especially when the periodic length of the road and the driving speed of the vehicle are combined to make the excitation frequency close to the low-order vibration mode of the vehicle body, the suspension support point may be in the response minimum area, resulting in a significant reduction in the displacement or air pressure change amplitude of the sensor output, while the actual wheel load is still changing normally. This speed-dependent response distortion cannot be identified or compensated in the traditional load estimation method, thus becoming an important problem restricting high-precision load estimation. SUMMARY

[0005] The present application provides a vehicle dynamic weighing and load estimation method, which solves the technical problems raised in the background art.

[0006] The present application provides a vehicle dynamic weighing and load estimation method, which includes:

[0007] The signal acquisition unit is configured to acquire a vehicle speed signal, a suspension tower top displacement signal and a road surface spatial pitch signal;

[0008] The road excitation processing unit is configured to calculate a main excitation frequency and an angular frequency based on the road surface spatial pitch signal in combination with the vehicle speed signal;

[0009] The modal and parameter unit is configured to perform modal analysis on the suspension tower top response to obtain a compliance frequency response function, extract a characteristic frequency, a bandwidth and an inverse frequency response function through parameterization processing, and construct a frequency weight function.

[0010] The synthetic observation unit is configured to generate a mode shape ratio based on the modal analysis result, and combine the suspension tower top displacement signal to form a synthetic observation signal.

[0011] The load estimation unit is configured to process the synthetic observation signal to obtain a load frequency domain estimation in combination with the inverse frequency response function, the main excitation angular frequency, the frequency weight and static parameters.

[0012] The net load unit is configured to perform signal conversion, filtering and constraint processing on the load frequency domain estimation to output a net load signal.

[0013] The present application has the beneficial effect that by introducing a dynamic correction mechanism based on the main excitation angular frequency and the inverse frequency response function under the condition of vehicle structure battery integration bottom plate, the adaptive compensation of the speed and road periodic excitation load estimation accuracy is realized. The present application can identify and correct the nonlinear deviation between the suspension displacement and the wheel load caused by the distortion of the structure dynamic response, so that the load estimation still maintains high precision and high stability under complex road and different vehicle speed conditions; at the same time, the frequency weight function is used to establish the cooperative relationship between the load estimation and the suspension control, and the net load signal is used to drive the suspension adaptive leveling process, realizing the integration of estimation and control. The present application can effectively improve the precision of vehicle dynamic weighing and attitude control, and significantly improve the smoothness and safety of the vehicle in actual driving. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 is a module diagram of the present application;

[0015] Figure 2 is a flowchart of the present application. DETAILED DESCRIPTION

[0016] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is merely meant to provide a better understanding of the subject matter described herein and can be changed in function and arrangement without departing from the scope of the present disclosure. Various processes or components can be omitted, substituted, or added according to desired implementations. Additionally, features described with respect to some examples can be combined in other examples.

[0017] like Figure 1 and Figure 2 As shown, a method for dynamic weighing and load estimation of vehicles includes:

[0018] The signal acquisition unit is used to acquire vehicle speed signals, suspension tower top displacement signals, and road surface spatial pitch signals.

[0019] The road surface excitation processing unit is used to calculate the main excitation frequency and angular frequency based on the road surface spatial pitch signal and the vehicle speed signal.

[0020] Modal and parametric units are used to perform modal analysis on the response at the top of the suspension tower to obtain the compliance frequency response function. After parameterization, characteristic frequencies, bandwidth, and inverse frequency response functions are extracted, and a frequency weighting function is constructed.

[0021] The synthetic observation unit is used to generate mode ratios based on modal analysis results and combine the displacement signals at the top of the suspension tower to form a synthetic observation signal.

[0022] The load estimation unit is used to process the synthetic observation signal by combining the inverse frequency response function, the main excitation angular frequency, the frequency weight and static parameters to obtain the load frequency domain estimate.

[0023] The net load unit is used to perform signal conversion, filtering, and constraint processing on the load frequency domain estimation to output the net load signal.

[0024] In one embodiment of the present invention, acquiring vehicle speed signal, suspension tower top displacement signal, and road surface spatial pitch signal includes:

[0025] S21, Set the sampling time to the kth time point, and the sampling period to a fixed time interval;

[0026] S22, when acquiring the vehicle speed signal, first obtain the cumulative pulse count at time k and time (k-1) through the wheel speed sensor, and calculate the difference between the cumulative pulse counts at the two times; then obtain the vehicle's equivalent rolling radius and the number of pulses per revolution of the wheel speed sensor; finally, multiply 2 by pi and then by the equivalent rolling radius, divide the result by the number of pulses per revolution, and then multiply by the ratio of the difference in the cumulative pulse counts to the sampling period to obtain the vehicle speed signal at time k.

[0027] S23, When acquiring the suspension tower top displacement signal, first acquire the original displacement signal of the left suspension tower top and the original displacement signal of the right suspension tower top;

[0028] S24, obtain the coefficient sequence of a low-pass filter of a preset order M, which contains coefficients from the 0th to the Mth.

[0029] S25, for the original displacement signal of the left side suspension tower top, the original displacement signal of the kth moment, the original displacement signal of the k-1th moment to the original displacement signal of the k-Mth moment are respectively multiplied by the 0th to the Mth coefficients in the low-pass filter coefficient sequence and then summed to obtain the left side suspension tower top displacement signal;

[0030] S26, the S25 is repeatedly executed for the original displacement signal of the right side suspension tower top to obtain the right side suspension tower top displacement signal;

[0031] S27, the road surface correlation signal is obtained, the front axle wheel center acceleration signal and the rear axle wheel center acceleration signal are obtained first; then the discrete cross-correlation function of the front axle wheel center acceleration signal and the rear axle wheel center acceleration signal is calculated, the discrete cross-correlation function is the product of the mth moment value of the front axle wheel center acceleration signal and the mth moment value of the rear axle wheel center acceleration signal, and the value of m ranges from 0 to M; then the corresponding τ value when the discrete cross-correlation function takes the maximum value is found as the time delay signal; finally, the kth moment speed signal is multiplied by the time delay signal to obtain the road surface space pitch signal.

[0032] The kth moment is the discrete time point identifier of signal collection (k is a positive integer, such as k=1, 2, 3...), which is used to distinguish the collection data at different times and ensure the time sequence correspondence.

[0033] The sampling period is the fixed time difference between adjacent two sampling moments, which is used to ensure the regularity of signal collection, and the common value is 0.01 to 0.1 seconds (selected according to the suspension response speed, and a small value is taken when the suspension dynamic response is fast).

[0034] The cumulative pulse number of the kth moment is the total pulse number recorded by the wheel speed sensor from the start of the vehicle to the kth sampling moment, which reflects the total number of turns of the tire and can be directly read from the counting module output by the wheel speed sensor.

[0035] The cumulative pulse number of the k-1th moment is the total pulse number recorded by the wheel speed sensor from the start of the vehicle to the k-1th sampling moment, which is used to calculate the pulse change in unit time in cooperation with the cumulative pulse number of the kth moment, and can be retrieved from the historical data of the sensor.

[0036] The difference value of the cumulative pulse number is the cumulative pulse number of the kth moment minus the cumulative pulse number of the k-1th moment, which reflects the pulse number generated by the tire rotation in the sampling period and is used to calculate the tire rotation speed; specifically, the difference value of the cumulative pulse number is equal to the cumulative pulse number of the kth moment minus the cumulative pulse number of the k-1th moment.

[0037] Equivalent rolling radius is the distance from the center of the tire to the ground when the tire is rolling under load (not the nominal radius of the tire, considering the deformation under load), which can be obtained by measurement (the vehicle drives a known distance, records the number of tire rotations, and divides the distance by the number of rotations and then by 2 and then by pi) or by consulting the load-rolling radius table provided by the tire manufacturer. The value for a common passenger car is 0.28 meters to 0.35 meters.

[0038] The number of pulses per revolution of the wheel speed sensor is the number of pulses output by the wheel speed sensor for each revolution of the tire, determined by the sensor hardware design (such as 100 pulses / revolution, 200 pulses / revolution), which can be read from the sensor technical manual or the vehicle ECU parameters.

[0039] The vehicle speed signal at the kth time is the instantaneous driving speed of the vehicle at the kth sampling time, with units of meters / second or kilometers / hour; specifically, the vehicle speed signal at the kth time is equal to (2 times pi times the equivalent rolling radius) divided by the number of pulses per revolution of the wheel speed sensor, then multiplied by (the difference in cumulative pulse count divided by the sampling period); if converted to kilometers / hour, multiply by 3.6.

[0040] The original displacement signal of the left side suspension tower top is the original position change signal of the left side suspension tower top (the connection point between the vehicle body and the suspension), which is directly collected by a displacement sensor (such as a laser displacement sensor, a linear potentiometer), containing noise and high-frequency interference components.

[0041] The original displacement signal of the right side suspension tower top is the original position change signal of the right side suspension tower top, collected in the same way as the left side, used to generate a combined observation signal with the left side signal.

[0042] The low-pass filter with a preset order of M is a linear filter used to filter out high-frequency noise in the suspension tower top original displacement signal, M is the order of the filter (commonly used values are 4 to 8 orders, the higher the order, the better the filtering effect but the larger the calculation amount), which needs to be set according to the noise frequency of the suspension displacement signal (such as 200 Hz and above for noise).

[0043] The coefficient sequence of the low-pass filter is the weight value to achieve the filtering function, the 0th to Mth coefficients correspond to the weighting proportion of the original signal at different times, which can be obtained by window function method (such as Hanning window, Hamming window), for example, the coefficient sequence of a 4th order Hanning window low-pass filter can be generated by the fir1 function of a signal processing software (such as MATLAB).

[0044] The left suspension tower top displacement signal is a left suspension tower top displacement signal filtered by a low-pass filter, high-frequency noise is removed, and the true displacement change is better reflected. Specifically, the left suspension tower top displacement signal is equal to (the left original displacement signal at the kth moment multiplied by the 0th coefficient of the low-pass filter) plus (the left original displacement signal at the k-1th moment multiplied by the 1st coefficient of the low-pass filter), and is sequentially accumulated until (the left original displacement signal at the k-Mth moment is multiplied by the Mth coefficient of the low-pass filter).

[0045] The right suspension tower top displacement signal is a right suspension tower top displacement signal filtered by a low-pass filter, and the processing mode is consistent with the left side. Specifically, the right suspension tower top displacement signal is equal to (the right original displacement signal at the kth moment multiplied by the 0th coefficient of the low-pass filter) plus (the right original displacement signal at the k-1th moment multiplied by the 1st coefficient of the low-pass filter), and is sequentially accumulated until (the right original displacement signal at the k-Mth moment is multiplied by the Mth coefficient of the low-pass filter).

[0046] The front axle wheel center acceleration signal is a vertical acceleration signal at the front axle wheel center, collected by an acceleration sensor installed near the front axle wheel center, and reflects the excitation of the road unevenness to the front axle.

[0047] The rear axle wheel center acceleration signal is a vertical acceleration signal at the rear axle wheel center.

[0048] The discrete cross-correlation function is an index for measuring the similarity between the front axle wheel center acceleration signal and the rear axle wheel center acceleration signal with different time delays, and the greater the value, the higher the similarity. Specifically, the discrete cross-correlation function is equal to (the 0th moment value of the front axle wheel center acceleration signal multiplied by the 0+τth moment value of the rear axle wheel center acceleration signal) plus (the 1st moment value of the front axle wheel center acceleration signal multiplied by the 1+τth moment value of the rear axle wheel center acceleration signal), and is sequentially accumulated until (the Mth moment value of the front axle wheel center acceleration signal is multiplied by the M+τth moment value of the rear axle wheel center acceleration signal).

[0049] m is the time index of the front axle wheel center acceleration signal, and the value range is 0 to M (M is consistent with the order of the low-pass filter), which ensures that the signal length participating in the cross-correlation calculation matches the filter order, avoiding inconsistent data lengths.

[0050] The value of τ is the time difference that makes the similarity between the front axle and the rear axle wheel center acceleration signal the highest, reflecting the time when the excitation received by the front axle wheel center is transmitted to the rear axle wheel center. The search range of τ is usually 0 to (sampling period multiplied by 10), avoiding exceeding the reasonable time delay range. Specifically, within the search range of τ (0 to the sampling period multiplied by 10), the discrete cross-correlation function value corresponding to each τ is calculated, and the τ corresponding to the maximum function value is selected as the time delay signal.

[0051] The road surface space pitch signal is the length of the road surface geometric repeating unit (such as the size of the Belgian blocks along the driving direction), and reflects the periodic characteristics of the road surface; specifically, the road surface space pitch signal is equal to the vehicle speed signal at the kth moment multiplied by the time delay signal.

[0052] In an embodiment of the present application, the main excitation frequency and the angular frequency are calculated based on the road surface space pitch signal combined with the vehicle speed signal, including:

[0053] The main excitation frequency signal is obtained by dividing the vehicle speed signal at the kth moment by the road surface space pitch signal;

[0054] The angular frequency signal is obtained by multiplying 2 by pi and then multiplying the main excitation frequency signal.

[0055] The main excitation frequency signal is the main frequency of the periodic excitation of the road surface acting on the vehicle suspension, with the unit of hertz, and directly determines the frequency domain interval of the suspension dynamic response; specifically, if the vehicle speed signal at the kth moment is in kilometers / hour, it needs to be converted to meters / second (conversion method: vehicle speed signal multiplied by 1000 and then divided by 3600), and then the converted vehicle speed signal is divided by the road surface space pitch signal (unit: meters) to obtain the main excitation frequency signal.

[0056] The angular frequency signal is an angular quantity representation of the main excitation frequency, with the unit of radian / second, and is used for subsequent frequency domain analysis (such as modal response calculation, inverse frequency response function application); specifically, the angular frequency signal is equal to 2 multiplied by pi (value 3.1416) and then multiplied by the main excitation frequency signal (unit: hertz).

[0057] In an embodiment of the present application, the modal analysis of the suspension tower top response is performed to obtain the compliance frequency response function, the characteristic frequency, the bandwidth and the inverse frequency response function are extracted through parameterization processing, and the frequency weight function is constructed, including:

[0058] The frequency spectrum of the suspension tower top displacement and the frequency spectrum of the external equivalent force are obtained, and the compliance frequency response function is obtained by dividing the frequency spectrum of the suspension tower top displacement by the frequency spectrum of the external equivalent force;

[0059] The amplitude of the compliance frequency response function is calculated, and the frequency corresponding to the minimum value of the amplitude is found as the characteristic frequency;

[0060] The minimum frequency deviation that makes the amplitude of the compliance frequency response function equal to twice the reference value of the amplitude of the compliance frequency response function at the characteristic frequency is found as the bandwidth;

[0061] The inverse frequency response function is obtained by dividing 1 by the compliance frequency response function;

[0062] When constructing the frequency weight function, first, the difference between the current frequency and the characteristic frequency is calculated, then the difference is divided by the bandwidth to obtain a square result, then 1 is added to the square result to obtain the denominator, and finally 1 is divided by the denominator to obtain the frequency weight function.

[0063] The frequency spectrum of the suspension tower top displacement is the amplitude-frequency distribution of the frequency domain converted from the suspension tower top time domain displacement signal (filtered displacement signal), which reflects the strength of the displacement response at different frequencies, and the unit is meter / hertz; specifically, first, the continuous 10-second suspension tower top time domain displacement signal (the sampling frequency is 1000 hertz, which ensures that the key frequency band of 20-30 hertz is covered) is collected; second, the time domain signal is processed by fast Fourier transform (FFT) to obtain the relationship between the frequency point and the corresponding displacement amplitude, that is, the frequency spectrum of the suspension tower top displacement.

[0064] The frequency spectrum of the external equivalent force is the amplitude-frequency distribution of the frequency domain converted from the time domain force signal acting on the suspension tower top, which reflects the strength of the excitation force at different frequencies, and the unit is newton / hertz; specifically, first, the suspension tower top is subjected to a sinusoidal sweep force (frequency range 5-50 hertz, covering the body modal frequency band) through a test bench (such as an electro-hydraulic servo test bench), or the time domain force signal during actual vehicle driving is collected through a force sensor installed on the suspension tower top; second, the time domain force signal is processed by fast Fourier transform (FFT) to obtain the relationship between the frequency point and the corresponding force amplitude, that is, the frequency spectrum of the external equivalent force.

[0065] The compliance frequency response function is a description of the correlation between the displacement response of the suspension tower top and the excitation force in the frequency domain under the excitation of the external equivalent force, which reflects the dynamic flexibility at different frequencies, and the unit is meter / newton; specifically, for each frequency point, the frequency spectrum amplitude of the suspension tower top displacement at the frequency is divided by the corresponding frequency spectrum amplitude of the external equivalent force, and the set of frequency points and corresponding ratios is the compliance frequency response function.

[0066] The amplitude of the compliance frequency response function is the modulus of the compliance frequency response function at each frequency point (only reflects the size, does not contain the phase), and the unit is meter / newton, which directly reflects the difficulty of generating displacement response under the excitation of the force (the smaller the amplitude, the smaller the dynamic flexibility, and the more difficult the displacement response).

[0067] The characteristic frequency is the frequency point with the smallest amplitude of the compliance frequency response function, which corresponds to the compliance zero point of the speed lock of the structure-battery integrated bottom plate vehicle, and the unit is hertz, which is the core reference frequency for subsequent load correction and suspension leveling; specifically, first, all frequency points (5-50 hertz) of the compliance frequency response function are traversed, and the amplitude corresponding to each frequency point is recorded; second, the frequency point with the smallest amplitude is selected, and the frequency is the characteristic frequency.

[0068] Bandwidth is the frequency interval width of the compliance zero point influence range of the velocity lock, in hertz, used to define the frequency band that needs to be corrected; specifically, first, determine the amplitude at the characteristic frequency as the reference value; second, find the first frequency point with an amplitude equal to twice the reference value on the left side of the characteristic frequency (lower frequency), and record it as the left boundary frequency; third, find the first frequency point with an amplitude equal to twice the reference value on the right side of the characteristic frequency (higher frequency), and record it as the right boundary frequency; fourth, subtract the left boundary frequency from the right boundary frequency, and the difference is the bandwidth.

[0069] Inverse frequency response function is the reciprocal of compliance frequency response function, reflecting the size of excitation force corresponding to displacement response at different frequencies, used to compensate for the load estimation deviation caused by the compliance zero point of the velocity lock, in N / m; specifically, for each frequency point, divide 1 by the compliance frequency response function value at that frequency, and the set of frequency points and corresponding results is the inverse frequency response function.

[0070] Current frequency is any frequency point traversed when constructing the frequency weight function, which needs to cover the compliance zero point influence interval of the velocity lock (characteristic frequency plus or minus 1.5 times the bandwidth), in hertz; specifically, the value range of the current frequency is set to the characteristic frequency minus 1.5 times the bandwidth, to the characteristic frequency plus 1.5 times the bandwidth, and discrete frequency points are taken at intervals of 0.1 hertz in this range, each discrete point is a current frequency.

[0071] Frequency weight function is a coefficient representing the strength of load correction demand at different frequencies, with a value range of 0-1 (the closer the current frequency is to the characteristic frequency, the closer the weight is to 1, and the stronger the correction demand); specifically, first, calculate the difference between the current frequency and the characteristic frequency; second, divide the difference by the bandwidth to get the square of the result; third, add 1 to the square result to get the denominator; fourth, divide 1 by the denominator to get the frequency weight function value corresponding to the current frequency.

[0072] In an embodiment of the present application, based on the modal analysis results, a mode shape ratio is generated, and a combined observation signal is formed by combining the suspension tower top displacement signals, including:

[0073] A set of discrete angular frequencies is determined, and the angular frequencies in the set satisfy that the difference between the current angular frequency and the characteristic frequency does not exceed the bandwidth;

[0074] When calculating the weighted cross-spectrum sum, in the set of discrete angular frequencies, the frequency weight function corresponding to each angular frequency, the conjugate of the left side suspension tower top displacement frequency spectrum, and the conjugate of the right side suspension tower top displacement frequency spectrum are multiplied, and all the products are summed to obtain the weighted cross-spectrum sum.

[0075] The weighted cross-spectrum sum is an index for measuring the correlation between the left and right suspension top displacement spectrums under the action of the frequency weight in the discrete angular frequency set, and the unit is (meter / hertz) square; specifically, the first step is to obtain the frequency weight function value, the left displacement spectrum value, and the conjugate value of the right displacement spectrum (if the right displacement spectrum is a complex number, the conjugate value is a number with the same real part and the opposite imaginary part) corresponding to each angular frequency in the discrete angular frequency set; the second step is to multiply the three values to obtain a product term of a single angular frequency; and the third step is to add the product terms of all angular frequencies, and the result is the weighted cross-spectrum sum.

[0076] When the synthetic observation signal spectrum is constructed, the product of the real part of the mode shape ratio and the left suspension top displacement spectrum is subtracted from the right suspension top displacement spectrum to obtain the synthetic observation signal spectrum; and the inverse discrete Fourier transform is performed on the synthetic observation signal spectrum to obtain the time-domain representation of the synthetic observation signal.

[0077] The discrete angular frequency set is a set of angular frequency points in the frequency domain affected by the compliance zero point of the focused velocity lock, and is used to limit the range of subsequent spectral analysis to avoid irrelevant frequency domain interference, and the unit is rad / s; specifically, the first step is to convert the characteristic frequency (Hz) into the corresponding angular frequency (characteristic angular frequency = 2 times pi times characteristic frequency), and to convert the bandwidth (Hz) into the corresponding angular frequency (bandwidth angular frequency = 2 times pi times bandwidth); the second step is to determine the angular frequency range as the characteristic angular frequency minus the bandwidth angular frequency to the characteristic angular frequency plus the bandwidth angular frequency; and the third step is to take the discrete angular frequency points in the range at an interval of 0.1 rad / s, and the set of all points is the discrete angular frequency set.

[0078] The weighted cross-spectrum sum is an index for measuring the correlation between the left and right suspension top displacement spectrums under the action of the frequency weight in the discrete angular frequency set, and the unit is (meter / hertz) square; specifically, the first step is to obtain the frequency weight function value, the left displacement spectrum value, and the conjugate value of the right displacement spectrum (if the right displacement spectrum is a complex number, the conjugate value is a number with the same real part and the opposite imaginary part) corresponding to each angular frequency in the discrete angular frequency set; the second step is to multiply the three values to obtain a product term of a single angular frequency; and the third step is to add the product terms of all angular frequencies, and the result is the weighted cross-spectrum sum.

[0079] The weighted cross-spectrum sum is an index for measuring the correlation between the left and right suspension top displacement spectrums under the action of the frequency weight in the discrete angular frequency set, and the unit is (meter / hertz) square; specifically, the first step is to obtain the frequency weight function value, the left displacement spectrum value, and the conjugate value of the right displacement spectrum (if the right displacement spectrum is a complex number, the conjugate value is a number with the same real part and the opposite imaginary part) corresponding to each angular frequency in the discrete angular frequency set; the second step is to multiply the three values to obtain a product term of a single angular frequency; and the third step is to add the product terms of all angular frequencies, and the result is the weighted cross-spectrum sum.

[0080] The mode shape ratio is a coefficient (characterizing the proportion and phase correlation of left and right displacements) reflecting the relationship between the left and right suspension top displacement modes in the discrete angular frequency set; specifically, the mode shape ratio is equal to the weighted cross-spectrum sum divided by the weighted self-spectrum sum.

[0081] The real part of the mode shape ratio is the real value part in the mode shape ratio complex form, used to eliminate the non-physical response component brought by the imaginary part.

[0082] The synthetic observation signal spectrum is a virtual displacement spectrum of the compliance zero point moving node interference of the speed lock, in units of meters / hertz, which can effectively avoid the response distortion caused by the original measurement point falling into the node; specifically, for each angular frequency in the discrete angular frequency set, subtract the right displacement spectrum value corresponding to the angular frequency from the left displacement spectrum value corresponding to the angular frequency (the real part of the mode shape ratio multiplied by the right displacement spectrum value corresponding to the angular frequency); the calculation result set of all angular frequencies is the synthetic observation signal spectrum.

[0083] The time domain representation of the synthetic observation signal is to convert the synthetic spectrum in the frequency domain into a displacement signal in the time dimension, in units of meters, which can be directly used for subsequent load estimation; specifically, first, determine the number of sampling points of the inverse discrete Fourier transform (consistent with the number of time domain sampling points when obtaining the suspension tower top displacement spectrum, such as 10000 points); second, process the synthetic observation signal spectrum using the inverse discrete Fourier transform algorithm; third, the output time-displacement data sequence is the time domain representation of the synthetic observation signal.

[0084] In an embodiment of the present application, the load frequency domain estimation is obtained by processing the synthetic observation signal in combination with the inverse frequency response function, the main excitation angular frequency, the frequency weight and the static parameters, including:

[0085] First, obtain the inverse frequency response function and the frequency weight function, the main excitation angular frequency, the synthetic observation signal spectrum, and the preset static stiffness parameter and the preset damping coefficient parameter;

[0086] For each frequency point corresponding to the main excitation angular frequency, first calculate the first coefficient by subtracting the frequency weight function from 1, then multiply the static stiffness parameter by the synthetic observation signal spectrum to obtain a first intermediate value, and add the product of the damping coefficient parameter and the frequency factor to obtain the first load component; then multiply the frequency weight function by the inverse frequency response function, and multiply the obtained result by the synthetic observation signal spectrum to obtain the second load component; add the first load component and the second load component to obtain the load frequency domain estimation value corresponding to the frequency point; traverse all frequency points corresponding to the main excitation angular frequency to complete the calculation of all load frequency domain estimation values, forming the load frequency domain estimation.

[0087] The static stiffness parameter is the stiffness value of the suspension under static load, reflecting the linear correlation between static displacement and static load, in units of N / m; specifically, it is obtained by static compression test on the test bench (applying gradient static force to the suspension, measuring the corresponding displacement, dividing the static force by the corresponding displacement, and taking the average value) or consulting the suspension design manual.

[0088] The damping coefficient parameter is an index of the ability of the suspension damping element to impede displacement changes, with units of newton-seconds / meter; specifically, it is obtained through a bench damping test (a sinusoidal displacement excitation is applied to the suspension, the corresponding damping force is measured, the damping force is divided by the displacement change rate, and the average value is taken) or by consulting a damping element technical manual.

[0089] The first coefficient is a weight coefficient for switching the static estimation mode, with a value range of 0-1 (the farther the frequency is from the compliance zero point of the speed lock, the closer the first coefficient is to 1, and the higher the proportion of the static estimation); specifically, the first coefficient is equal to 1 minus the frequency weight function value of the current main excitation angular frequency corresponding frequency point.

[0090] The frequency factor is a frequency-related quantity representing the dynamic damping effect, with a value equal to the current main excitation angular frequency (since the damping force is proportional to the displacement change rate, the angular frequency directly reflects the change rate characteristic), with units of radians / second; specifically, the frequency factor is equal to the angular frequency value corresponding to the current main excitation angular frequency.

[0091] The first intermediate value is a load base quantity under the joint action of static stiffness and dynamic damping, with units of newtons; specifically, the first intermediate value is equal to (the static stiffness parameter multiplied by the synthesized observation signal spectrum value of the current frequency point) plus (the damping coefficient parameter multiplied by the frequency factor).

[0092] The first load component is a load estimation part based on static stiffness and damping, mainly acting on the compliance zero point interval of the speed lock, with units of newtons; specifically, the first load component is equal to the first coefficient multiplied by the first intermediate value.

[0093] The second load component is a dynamic correction load part based on the inverse frequency response function, mainly acting on the compliance zero point interval of the speed lock, used to compensate for estimation deviation, with units of newtons; specifically, the second load component is equal to (the frequency weight function value of the current frequency point multiplied by the inverse frequency response function value of the frequency point) multiplied by the synthesized observation signal spectrum value of the frequency point.

[0094] The load frequency domain estimation value is the final load estimation result of the frequency point corresponding to a single main excitation angular frequency, with units of newtons; specifically, the load frequency domain estimation value is equal to the first load component plus the second load component.

[0095] The load frequency domain estimation is a collection of load frequency domain estimation values of frequency points corresponding to all main excitation angular frequencies, reflecting the load distribution at different frequencies, with units of newtons; specifically, first, determine the frequency point range corresponding to the main excitation angular frequency (consistent with the frequency range corresponding to the discrete angular frequency set); second, calculate the load frequency domain estimation value for each frequency point according to the above steps; third, the combination of all frequency points and corresponding estimation values is the load frequency domain estimation.

[0096] In one embodiment of the present invention, signal conversion, filtering, and constraint processing are performed on the load frequency domain estimation to output a net load signal, including:

[0097] To obtain the load frequency domain estimate, determine the discrete angular frequency set and sampling time, and count the total number of angular frequencies in the discrete angular frequency set. For the load frequency domain estimate corresponding to each angular frequency in the set, extract the real and imaginary parts respectively. Then calculate the product of each angular frequency and the current sampling time, and obtain the cosine and sine values ​​corresponding to the product respectively. Next, multiply the real part of each load frequency domain estimate by the cosine value of the product of the corresponding angular frequencies, multiply the imaginary part of each load frequency domain estimate by the sine value of the product of the corresponding angular frequencies, and subtract the sine value from the cosine value to obtain the calculated value corresponding to each angular frequency. Finally, sum the calculated values ​​corresponding to all angular frequencies, and divide the sum by the total number of angular frequencies in the discrete angular frequency set to obtain the load time domain estimate.

[0098] Obtain a low-pass filter of a preset order and its corresponding coefficient sequence, which includes coefficients from the 0th to the coefficients corresponding to the filter order; take the load time-domain estimate at the current sampling time, and the load time-domain estimate from the previous time to the time corresponding to the current sampling time minus the filter order; multiply the load time-domain estimate by each of the coefficients from the 0th to the coefficients corresponding to the filter order in the low-pass filter coefficient sequence, and then sum all the products to obtain the low-pass load time-domain estimate;

[0099] Obtain the vehicle mass parameters and gravitational acceleration parameters. Multiply the vehicle mass parameters by the gravitational acceleration parameters to obtain the total gravity. Then, count all corner points of the vehicle and calculate the sum of the time-domain estimates of the low-pass load corresponding to each corner point. Next, divide the total gravity by this sum to obtain the scaling factor. Finally, multiply the time-domain estimate of the low-pass load corresponding to each corner point by the scaling factor to obtain the net load signal corresponding to each corner point.

[0100] The sampling time is the discrete time point at which the load time-domain transformation is currently performed, relative to the k-th time point, and is measured in seconds; specifically, the sampling time is equal to the index of the k-th sampling time point multiplied by the sampling period duration.

[0101] The total number of angular frequencies is the number of angular frequency points contained in the discrete angular frequency set, and it has no unit; specifically, the total number of angular frequencies is equal to the result of counting all angular frequency points in the discrete angular frequency set.

[0102] The real part of the load frequency domain estimation is the real numerical part of the complex form of the load frequency domain estimation, reflecting the amplitude contribution of the load frequency domain components, and is measured in Newtons.

[0103] The imaginary part of the load frequency domain estimation is the imaginary numerical part in the complex form of the load frequency domain estimation, reflecting the phase contribution of the load frequency domain components, and is measured in Newtons.

[0104] The product of the angular frequency and the current sampling time is a phase-related item of the conversion of the frequency domain signal to the time domain signal, and has a unit of radian; specifically, the product of the angular frequency and the current sampling time is equal to each angular frequency value multiplied by the current sampling time value. The cosine value corresponding to the product is a cosine function result of the phase item, and has no unit; and is obtained through a trigonometric function calculation tool or algorithm (such as a trigonometric function library in a vehicle-mounted controller). The sine value corresponding to the product is a sine function result of the phase item, and has no unit; and is obtained through a trigonometric function calculation tool or algorithm (such as a trigonometric function library in a vehicle-mounted controller).

[0105] The calculated value corresponding to the angular frequency is a contribution of a single angular frequency to the time domain load, and has a unit of newton; specifically, the calculated value corresponding to each angular frequency is equal to (the real part of the load frequency domain estimation of the angular frequency multiplied by the cosine value of the corresponding product) minus (the imaginary part of the load frequency domain estimation of the angular frequency multiplied by the sine value of the corresponding product).

[0106] The load time domain estimation is a load signal converted from the frequency domain load to the time dimension, and has a unit of newton; specifically, the load time domain estimation is equal to the sum of the calculated values corresponding to all angular frequencies divided by the total number of angular frequencies.

[0107] The low-pass filter of the preset order is a filter for filtering out high-frequency dynamic components in the load time domain estimation, and has an order of 4-8 orders, so as to ensure that the narrow-band kinetic energy corresponding to the compliance zero point of the speed lock is filtered out; specifically, the low-pass filter is designed through a window function method (such as a Hanning window) and generated in combination with a signal processing software (such as MATLAB).

[0108] The load time domain estimation after low-pass filtering is a stationary load signal filtered out of high-frequency interference, and has a unit of newton, and is used for subsequent net load calculation; specifically, the load time domain estimation after low-pass filtering is equal to (the load time domain estimation at the current sampling time multiplied by the 0th coefficient of the low-pass filter) plus (the load time domain estimation at the time point before the current sampling time multiplied by the 1st coefficient of the low-pass filter), and is sequentially added until (the load time domain estimation at the time point subtracted by the filter order from the current sampling time multiplied by the Mth coefficient of the low-pass filter, M being the filter order).

[0109] The vehicle mass parameter is the total mass of the sprung part of the vehicle (including the vehicle body, the battery, the occupants, etc.), and has a unit of kilogram; and is obtained through a vehicle weighing test (such as a ground scale weighing) or by referring to the sprung mass parameter in a vehicle design manual.

[0110] The gravity acceleration parameter is a standard value of the gravity acceleration on the surface of the earth, and has a unit of meter per square second; and has a value of 9.8 (an industry general standard value, meeting the vehicle-mounted control precision requirement).

[0111] The total gravity is the total weight of the sprung part of the vehicle, in units of newtons, used to constrain the physical reasonableness of the load estimate; specifically, the total gravity is equal to the vehicle mass parameter multiplied by the gravity acceleration parameter (9.8 meters per second squared).

[0112] The sum of the low-passed load time-domain estimates of all the corner points is the cumulative value of the low-passed loads of the four corner points (front axle left / right, rear axle left / right) of the vehicle, in units of newtons; specifically, the sum is equal to the low-passed load time-domain estimate of the front axle left corner point plus the low-passed load time-domain estimate of the front axle right corner point, plus the low-passed load time-domain estimate of the rear axle left corner point, and finally plus the low-passed load time-domain estimate of the rear axle right corner point.

[0113] The scale factor is a coefficient used to correct the load estimation deviation and ensure matching with the total gravity, and has no unit; specifically, the scale factor is equal to the total gravity divided by the sum of the low-passed load time-domain estimates of all the corner points.

[0114] The net load signal is the accurate load signal finally used for suspension adaptive leveling, in units of newtons; specifically, the net load signal of each corner point is equal to the low-passed load time-domain estimate of the corner point multiplied by the scale factor.

[0115] The above describes the embodiments of the embodiments, but the embodiments are not limited to the specific embodiments described above, and the specific embodiments described above are only illustrative and not limiting, and those of ordinary skill in the art can make many forms under the inspiration of the embodiments, all of which belong to the protection of the embodiments.

Claims

1. A method for dynamic weighing and load estimation of vehicles, characterized in that, include: Acquire vehicle speed signal, suspension tower top displacement signal, and road surface pitch signal; The main excitation frequency and angular frequency are calculated based on the road space pitch signal combined with the vehicle speed signal. Modal analysis was performed on the response at the top of the suspension tower to obtain the compliance frequency response function. After parameterization, the characteristic frequency, bandwidth and inverse frequency response function were extracted, and a frequency weighting function was constructed. Based on the modal analysis results, the mode shape ratio is generated, and the displacement signals at the top of the suspension tower are combined to form a synthetic observation signal; By combining the inverse frequency response function, the main excitation angular frequency, the frequency weight, and the static parameters, the load frequency domain estimate is obtained by processing the synthetic observation signal. The load frequency domain estimation is subjected to signal conversion, filtering and constraint processing to output the net load signal.

2. The vehicle dynamic weighing and load estimation method according to claim 1, characterized in that, Acquire vehicle speed signal, suspension tower top displacement signal, and road surface pitch signal, including: S21, Set the sampling time to the kth time point, and the sampling period to a fixed time interval; S22, when acquiring the vehicle speed signal, first obtain the cumulative pulse count at time k and time (k-1) through the wheel speed sensor, and calculate the difference between the cumulative pulse counts at the two times; then obtain the vehicle's equivalent rolling radius and the number of pulses per revolution of the wheel speed sensor; finally, multiply 2 by pi and then by the equivalent rolling radius, divide the result by the number of pulses per revolution, and then multiply by the ratio of the difference in the cumulative pulse counts to the sampling period to obtain the vehicle speed signal at time k. S23, When acquiring the suspension tower top displacement signal, first acquire the original displacement signal of the left suspension tower top and the original displacement signal of the right suspension tower top; S24, obtain the coefficient sequence of a low-pass filter of a preset order M, which contains coefficients from the 0th to the Mth. S25, For the original displacement signal of the left suspension tower top, the original displacement signal at time k, the original displacement signal at time k-1 up to time kM are multiplied one by one with the coefficients from the 0th to the Mth in the low-pass filter coefficient sequence and then summed to obtain the displacement signal of the left suspension tower top. S26, Repeat S25 for the original displacement signal of the right suspension tower top to obtain the right suspension tower top displacement signal; S27. Obtain road surface correlation signals. First, obtain the acceleration signals of the front axle wheel center and the rear axle wheel center. Then, calculate the discrete cross-correlation function of the acceleration signals of the front axle wheel center and the rear axle wheel center. This discrete cross-correlation function is the sum of the product of the m-th time value of the front axle wheel center acceleration signal and the m-th plus τ-th time value of the rear axle wheel center acceleration signal, where m ranges from 0 to M. Then, find the τ value corresponding to the maximum value of the discrete cross-correlation function, and use it as the time delay signal. Finally, multiply the vehicle speed signal at the k-th time by this time delay signal to obtain the road surface spatial pitch signal.

3. The vehicle dynamic weighing and load estimation method according to claim 2, characterized in that, The main excitation frequency and angular frequency are calculated based on the road surface pitch signal combined with the vehicle speed signal, including: Divide the vehicle speed signal at time k by the road space pitch signal to obtain the main excitation frequency signal; Multiply 2 by pi and then by the main excitation frequency signal to obtain the angular frequency signal.

4. The vehicle dynamic weighing and load estimation method according to claim 3, characterized in that, Modal analysis of the suspension tower top response yields the compliance frequency response function. After parameterization, characteristic frequencies, bandwidth, and inverse frequency response functions are extracted. A frequency weighting function is then constructed, including: Obtain the spectrum of the suspension tower top displacement and the spectrum of the external equivalent force. Divide the spectrum of the suspension tower top displacement by the spectrum of the external equivalent force to obtain the compliance frequency response function. Calculate the amplitude of the compliance frequency response function, find the frequency at which the amplitude reaches its minimum value, and use it as the characteristic frequency; Using the amplitude of the compliance frequency response function at the characteristic frequency as the reference value, find the minimum frequency deviation that makes the amplitude of the compliance frequency response function equal to twice the reference value, and use it as the bandwidth; Divide 1 by the compliance frequency response function to obtain the inverse frequency response function; When constructing the frequency weighting function, first calculate the difference between the current frequency and the characteristic frequency, then divide the difference by the bandwidth and square it to get the squared result. Then add 1 to the squared result to get the denominator, and finally divide 1 by the denominator to get the frequency weighting function.

5. The vehicle dynamic weighing and load estimation method according to claim 4, characterized in that, Based on the modal analysis results, a mode shape ratio is generated, and the displacement signals at the top of the suspension tower are combined to form a synthetic observation signal, including: Determine a set of discrete angular frequencies, wherein the angular frequencies in the set satisfy the condition that the difference between the current angular frequency and the characteristic frequency does not exceed the bandwidth; When calculating the weighted cross spectrum sum, within the discrete angular frequency set, the frequency weight function corresponding to each angular frequency, the left suspension tower top displacement spectrum, and the right suspension tower top displacement spectrum are multiplied by their conjugates, and then all products are summed to obtain the weighted cross spectrum sum. When calculating the weighted self-spectral sum, within the discrete angular frequency set, the frequency weight function corresponding to each angular frequency is multiplied by the square of the modulus of the displacement spectrum of the right suspension tower top, and then all products are summed to obtain the weighted self-spectral sum; the weighted cross-spectral sum is divided by the weighted self-spectral sum to obtain the mode ratio; the real part of the mode ratio is taken. When constructing the spectrum of the synthetic observation signal, the product of the real part of the mode ratio and the spectrum of the displacement of the right suspension tower top is subtracted from the spectrum of the displacement of the left suspension tower top to obtain the spectrum of the synthetic observation signal; the time domain representation of the synthetic observation signal is obtained by performing an inverse discrete Fourier transform on the spectrum of the synthetic observation signal.

6. The vehicle dynamic weighing and load estimation method according to claim 5, characterized in that, By combining the inverse frequency response function, the main excitation angular frequency, frequency weights, and static parameters, the composite observation signal is processed to obtain a load frequency domain estimate, including: First, obtain the inverse frequency response function and frequency weighting function, the main excitation angular frequency, the spectrum of the synthesized observation signal, and the preset static stiffness parameters and preset damping coefficient parameters; For each frequency point corresponding to the main excitation angular frequency, first calculate 1 minus the frequency weighting function to obtain the first coefficient, then multiply the static stiffness parameter by the spectrum of the synthesized observed signal, and add the product of the damping coefficient parameter and the frequency factor to obtain the first intermediate value. Multiply the first coefficient and the first intermediate value to obtain the first load component. Then multiply the frequency weighting function by the inverse frequency response function, and multiply the result by the spectrum of the synthesized observed signal to obtain the second load component. Add the first load component and the second load component to obtain the load frequency domain estimate of the corresponding frequency point. Iterate through all frequency points corresponding to the main excitation angular frequencies to complete the calculation of all load frequency domain estimates, forming the load frequency domain estimate.

7. The method for dynamic weighing and load estimation of a vehicle according to claim 6, characterized in that, The load frequency domain estimation is subjected to signal conversion, filtering, and constraint processing to output the net load signal, including: To obtain the load frequency domain estimate, determine the discrete angular frequency set and sampling time, and count the total number of angular frequencies in the discrete angular frequency set. For the load frequency domain estimate corresponding to each angular frequency in the set, extract the real and imaginary parts respectively. Then calculate the product of each angular frequency and the current sampling time, and obtain the cosine and sine values ​​corresponding to the product respectively. Next, multiply the real part of each load frequency domain estimate by the cosine value of the product of the corresponding angular frequencies, multiply the imaginary part of each load frequency domain estimate by the sine value of the product of the corresponding angular frequencies, and subtract the sine value from the cosine value to obtain the calculated value corresponding to each angular frequency. Finally, sum the calculated values ​​corresponding to all angular frequencies, and divide the sum by the total number of angular frequencies in the discrete angular frequency set to obtain the load time domain estimate. Obtain a low-pass filter of a preset order and its corresponding coefficient sequence, which includes coefficients from the 0th to the coefficients corresponding to the filter order; take the load time-domain estimate at the current sampling time, and the load time-domain estimate from the previous time to the time corresponding to the current sampling time minus the filter order; multiply the load time-domain estimate by each of the coefficients from the 0th to the coefficients corresponding to the filter order in the low-pass filter coefficient sequence, and then sum all the products to obtain the low-pass load time-domain estimate; Obtain the vehicle mass parameters and gravitational acceleration parameters. Multiply the vehicle mass parameters by the gravitational acceleration parameters to obtain the total gravity. Then, count all corner points of the vehicle and calculate the sum of the time-domain estimates of the low-pass load corresponding to each corner point. Next, divide the total gravity by this sum to obtain the scaling factor. Finally, multiply the time-domain estimate of the low-pass load corresponding to each corner point by the scaling factor to obtain the net load signal corresponding to each corner point.

Images