A method for real-time estimation of vehicle mass

Abstract

Aiming at the situation where the vehicle mass changes frequently and there are measurement errors in the observable quantities, with the goal of improving the rapidity and stability of the vehicle mass estimation process, a vehicle mass estimation method based on the Dynamic Forgetting Factor Recursive Least Squares Method (hereinafter referred to as DFFRLS) is proposed. According to the vehicle longitudinal dynamics principle, this method determines the system output and observable quantities of the Recursive Least Squares Method (RLS) for vehicle mass estimation; and by analyzing the influence of the forgetting factor on the estimation result of the recursive least squares method and the variation laws of vehicle operating conditions and vehicle mass, the forgetting factor is improved to a dynamically changing mode associated with vehicle speed and stop time points. A 10.5-meter vehicle simulation test environment is established using Matlab / Simulink to conduct a comparative verification of vehicle mass estimation between FFRLS and DFFRLS. The estimation time of DFFRLS is within 2 s, and the volatility of the vehicle mass estimation value is better than that of FFRLS. The rapidity and stability of vehicle mass estimation by DFFRLS are both better than those of FFRLS. It shows that DFFRLS can estimate the vehicle mass quickly and stably.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle control, and particularly relates to a method for real-time estimation of vehicle mass. Background Art

[0002] At present, great progress has been made in the research on vehicle mass estimation based on the recursive least squares method and its improved methods. However, for the situation where the vehicle mass of buses changes frequently and there are measurement errors in the observed quantities, there are relatively few studies that can balance the rapidity and stability of the bus vehicle mass estimation process.

[0003] Vehicle mass is an important parameter for intelligent vehicle control and decision-making. As the main source of rolling resistance, ramp resistance, etc., it will affect the shift strategy of the automatic transmission, and thus affect the power performance and economy of the vehicle. Combining sensing measurement technology and parameter estimation algorithms to accurately obtain the vehicle mass in real time is of great significance and application value for vehicle driving decisions. Currently, the common methods for vehicle mass estimation are the recursive least squares method and its improved methods.

[0004] In the literature "Estimation of Vehicle Mass and Road Grade Considering Lateral Motion [J / OL]", the vehicle dynamics model is corrected by using acceleration, and the least squares algorithm with a forgetting factor is used to enhance the adaptability of new data to the time-varying characteristics of the vehicle system to estimate the vehicle mass; in the literature "Joint Estimation of Freight Vehicle Mass and Road Grade under Complex Conditions [J / OL]", aiming at the stability problem of mass estimation during the acceleration process of freight vehicles, the influence of different acceleration types on mass estimation is explored, and a joint estimation method of M-estimation and recursive least squares method based on a forgetting factor is proposed to achieve robust estimation of freight vehicle mass; in the literature "Electric Vehicle Ramp Identification by Integrating Vehicle Mass Estimation [J / OL]", by establishing a vehicle longitudinal dynamics model, analyzing the signal characteristics of the speed sensor under actual vehicle conditions, a least squares vehicle mass estimation strategy with a forgetting factor is constructed to directly obtain the vehicle mass under the starting condition; in the literature "Vehicle Mass Estimation Based on Longitudinal Frequency Response Characteristics [J]", based on the obtained longitudinal acceleration and wheel speed information, and based on the vehicle longitudinal dynamics model, the amplitude-frequency function relationship between the acceleration and the wheel speed signal is deduced, and the amplitude ratio between the acceleration at the vehicle centroid and the wheel speed at different frequencies is obtained, and then the vehicle mass is obtained by fitting with the least squares method; in the literature "Vehicle Mass and Road Grade Estimation Based on Interactive Multiple Models [J]", the working conditions suitable for accurate mass estimation are set, and accordingly, an algorithm for calculating the confidence factor of mass estimation based on fuzzy rules is proposed, and then a recursive least squares vehicle mass estimation algorithm based on the confidence factor is designed to achieve online mass estimation. Summary of the Invention

[0005] The object of the present invention is to propose a real-time vehicle mass estimation method based on a dynamic forgetting factor in view of the problems existing in the prior art, which aims to solve the problems such as the rapidity and stability of vehicle mass estimation in the existing vehicle mass estimation process.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A real-time vehicle mass estimation method, comprising the following steps:

[0008] Construct a vehicle longitudinal dynamics model:

[0009] T dr / r tire -(ρ·A·C w ·v^2) / 2 - ε·mg·cos(α) - mg·sin(α) = ma

[0010] In the formula, vehicle operation data is collected; the determined parameters are rtire, the radius of the vehicle tire, unit m, ρ, the air density, taken as 1.29 kg / m^3, A, the frontal area of the measured vehicle, unit m^2, CW, the wind resistance coefficient of the test environment, ε, the rolling resistance coefficient of the measured vehicle, g, the gravitational acceleration, taken as 9.8 N / kg, α, the real-time collected slope, taken as 0 to 1, v, the vehicle speed, unit m / s, a, the acceleration, unit m / s^2, and the driving torque T dr .

[0011] Preferably, when the vehicle longitudinal dynamics model is converted into the following form, it is convenient for mass calculation:

[0012] (T dr / r tire -(ρ·A·C w ·v^2) / 2) = m([a + ε·g·cos(α) + g·sin(α)])

[0013] In this formula, the data to be solved in real time is the mass of the measured vehicle, and the known parameters are the vehicle tire radius, air density, frontal area of the measured vehicle, wind resistance coefficient of the test environment, rolling resistance coefficient of the measured vehicle, and gravitational acceleration; the parameters collected in real time are slope, vehicle speed, acceleration, and driving torque.

[0014] Preferably, the measurement errors of vehicle speed, acceleration, and driving torque are simulated by introducing white noise to increase the robustness of the experiment.

[0015] Construct a vehicle mass model regarding the least squares method:

[0016]

[0017] The update formula of the increment matrix is:

[0018]

[0019] The covariance matrix update formula is as follows:

[0020]

[0021] In the formula, M(k) represents the estimation quality at the k-th step, K(k) represents the gain at the k-th step, E F (k) represents the output of the dynamic equation at the k-th step, E α (k) represents the observation at the k-th step, k represents the k-th step, M(k - 1) represents the quality at the (k - 1)-th step, the model P(k - 1) represents the covariance at the (k - 1)-th step, P(k) represents the covariance at the k-th step, and λ represents the forgetting factor.

[0022] When k = 0, the following values are taken:

[0023]

[0024] α is a sufficiently large positive real number, ε is a zero vector or a sufficiently small positive real vector. E is the identity matrix.

[0025] According to the longitudinal dynamics analysis of the vehicle, the observable E α of the system can be obtained:

[0026] E α = α + ε·g·cos(·α) + g·sin(·α)

[0027] According to the longitudinal dynamics analysis of the vehicle, the output E F of the system can be obtained:

[0028]

[0029] Determine the value range and change trend of the dynamic forgetting factor.

[0030] Preferably, when the value range of the forgetting factor is 0.8 - 1, the stability and real-time performance of the quality calculation are relatively high. At the beginning of the calculation, it is set to 0.8; as the system running time increases, the forgetting factor approaches 1;

[0031] According to the value range and change trend of the dynamic forgetting factor, use the exponential increasing formula to express the calculation formula of the dynamic forgetting factor, and obtain:

[0032]

[0033] Among them, Δt is the iteration period, which is 0.1s; k0 is the iteration number when the vehicle restarts and drives;

[0034] Try to find the appropriate values of the dynamic factor parameters A, B, and C according to the exhaustive method;

[0035] Bring the dynamic forgetting factor into the least squares method model for mass to obtain the mass at time k.

[0036] The present invention also provides a system for assisting vehicle mass estimation, which is characterized by including the following modules:

[0037] A Vehcle module for calculating vehicle speed, acceleration, friction, combined torque, and slope;

[0038] A Mass_RLS module for estimating vehicle mass;

[0039] The Mass_RLS module adopts any one of the above vehicle mass real-time estimation methods, and the Mass_RLS module includes:

[0040] A lam module for calculating the dynamic forgetting factor;

[0041] An EF_Cal module for calculating system observables;

[0042] An EA_Cal module for calculating system output;

[0043] An FFRLS module for the recursive least squares method;

[0044] The FFRLS module includes:

[0045] An FFRLS_Init module for recording parameters at time K - 1.

[0046] The present invention also provides a device for real-time mass estimation on a vehicle, including the following modules: a data acquisition module, a device for collecting vehicle operation data; a data transmission module, which transmits the collected data into the CAN network in the Matlab algorithm system; a mass calculation module, which receives the data in the CAN network and transmits it into Matlab, and uses the above system for assisting vehicle mass estimation to estimate the mass.

[0047] Traditional vehicle mass estimation methods mostly rely on fixed parameter models and are easily affected by changes in working conditions, such as road conditions and load changes, resulting in an increase in mass estimation errors. However, by introducing a dynamic forgetting factor, the present invention can adaptively adjust the parameter update rate of the model according to the real-time vehicle operation state, thereby improving the stability and real-time performance of mass estimation. Brief Description of the Drawings

[0048] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention.

[0049] Figure 1For the driving conditions of CCBC buses

[0050] Figure 2 For the vehicle weight change under set conditions

[0051] Figure 3 For the estimated vehicle mass with a forgetting factor of 0.80

[0052] Figure 4 For the estimated vehicle mass with a forgetting factor of 0.95

[0053] Figure 5 For the estimated vehicle mass with a forgetting factor of 0.99

[0054] Figure 6 For the estimated vehicle mass with a dynamic forgetting factor

[0055] Figure 7 For the algorithm module diagram

[0056] Figure 8 For the structural diagram of the simulation bus parameter calculation algorithm

[0057] Figure 9 For the system output E F Calculation algorithm structural diagram

[0058] Figure 10 For the system observable E α Calculation algorithm structural diagram

[0059] Figure 11 For the structural diagram of the dynamic forgetting factor calculation algorithm

[0060] Figure 12 For the structural diagram of the vehicle mass calculation algorithm

[0061] Figure 13 For the structural diagram of the FFRLS forgetting factor recursive least squares algorithm

[0062] Figure 14 For the structural diagram of the FFRLS_Init module algorithm

[0063] Figure 15 For the process schematic diagram of the vehicle mass measurement device Specific implementation mode

[0064] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, so they are only examples and cannot be used to limit the protection scope of the present invention.

[0065] Unless otherwise specified, the technical terms or scientific terms used in this application shall have the ordinary meanings understood by those skilled in the art. The terms "first", "second", etc. used in this specification, claims and drawings are used to distinguish similar objects and do not indicate a specific order or priority. It is understood that these data can be interchanged under appropriate circumstances to implement the vehicle mass real-time estimation method described in the embodiments of the present invention. In addition, the terms "comprising" and "having" and any variations thereof are intended to mean non-exclusive inclusion. Unless otherwise specified, the term "plurality" represents two or more possibilities. In the embodiments of the present invention, the character " / " is used to indicate an "or" relationship between the front and rear objects. For example, sensor A / sensor B means sensor A or sensor B. The term "and / or" is used to describe the association relationship of objects and can represent three relationships. For example, sensor A and / or sensor B means sensor A, sensor B, or sensor A and sensor B exist simultaneously. The term "corresponding" refers to an association relationship or binding relationship. For example, vehicle load data corresponds to a vehicle mass estimation model, which means that there is an association or binding relationship between the two.

[0066] This embodiment provides an example of bus mass estimation for a vehicle mass real-time estimation method, and the steps include:

[0067] Construct a vehicle longitudinal dynamics model:

[0068] (T dr / r tire -(ρ·A·C w ·v^2) / 2)=m([a+ε·g·cos(α)+g·sin(α)])

[0069] In this formula, the data to be solved in real time is the mass of the measured vehicle, and the known parameters are the vehicle tire radius, air density, the frontal area of the measured vehicle, the wind resistance coefficient of the test environment, the rolling resistance coefficient of the measured vehicle, and the gravitational acceleration. The parameters collected in real time are the slope, vehicle speed, acceleration, and driving torque.

[0070] Under the Matlab / Simulink environment, build a 10.5-meter bus vehicle model to run the CCBC (China Typical City Bus Cycle) condition. The main parameters of the bus are as follows

[0071]

[0072] Simulated bus parameters

[0073] The driving situation of the bus is as Figure 1 ;

[0074] Eliminate the influence of the slope on vehicle mass estimation, and set to drive on a flat road with a slope of zero.

[0075] Set the driving torque according to the vehicle speed, acceleration, and slope settings under the CCBC condition, as well as the vehicle longitudinal dynamics formula:

[0076]

[0077] The maximum carrying capacity of the bus is 90 people, and the average mass per person is 70 kg (including the passenger's weight and carry-on luggage). The change in vehicle weight is as shown in Figure 2 .

[0078] Simulate the measurement errors of vehicle speed, acceleration, and driving torque by introducing white noise. In this example, the noise power of the vehicle speed white noise is set to 0.001, and the sampling time is 0.01 s; the noise power of the acceleration white noise is set to 0.000001, and the sampling time is 0.01 s; the noise power of the driving torque white noise is set to 0.01, and the sampling time is 0.02 s.

[0079] Construct a vehicle mass calculation method based on the forgetting factor least squares method:

[0080]

[0081] In the formula, M(k) represents the estimated mass at the k-th step, K(k) represents the gain at the k-th step, E F (k) represents the output of the dynamic equation at the k-th step, E α (k) represents the observation at the k-th step, k represents the k-th step, and M(k - 1) represents the mass at the (k - 1)-th step;

[0082] The update formula for the incremental matrix is:

[0083]

[0084] The update formula for the covariance matrix is:

[0085]

[0086] P(k - 1) represents the covariance at the (k - 1)-th step, P(k) represents the covariance at the k-th step, and λ represents the forgetting factor.

[0087] When k = 0, the following values are taken:

[0088]

[0089] α is a positive real number that is sufficiently large, ε is a zero vector or a positive real vector that is sufficiently small, and E is the identity matrix.

[0090] According to the vehicle longitudinal dynamics analysis, the model of the system observable E α can be obtained:

[0091] E α= α + ε·g·cos(·α) + g·sin(·α)

[0092] According to the longitudinal dynamics analysis of the vehicle, the system output E F model is as follows:

[0093]

[0094] Determine the value range of the dynamic forgetting factor. The value range of the forgetting factor is generally 0.8 - 1 (not taking 1). When initially estimating, it is set to 0.8. As the system running time increases, the forgetting factor approaches 1.

[0095] According to the setting method of the value range and change trend of the forgetting factor, use the exponential increasing formula to express the calculation formula of the dynamic forgetting factor, and obtain:

[0096]

[0097] where, Δt is the iteration period, which is 0.1s; k0 is the iteration number when the vehicle restarts and runs;

[0098] Try to find the appropriate values of the dynamic factor parameters A, B, and C according to the exhaustive method;

[0099] The formula obtained in this example is:

[0100]

[0101] Substitute the dynamic forgetting factor into the least squares model regarding mass to obtain the mass at time k.

[0102] The present invention can achieve fast and stable vehicle weight estimation. The vehicle weight estimation results with forgetting factors of 0.8, 0.95, 0.99 and the dynamic form are compared, and the comparison results are as Figure 3 Figure 4 Figure 5 Figure 6 .

[0103] Regarding the rapidity of the bus vehicle mass estimation process, set the time from when the bus starts running from the stop to when the difference between the vehicle mass estimation value and the true value is within 1% as the rapidity measurement standard.

[0104] Regarding the stability of the bus vehicle mass estimation process, set V as the estimation value volatility after the error between the bus vehicle mass estimation value and the true value is less than 1%, with the unit of (kg / m), and the calculation method is as follows:

[0105]

[0106] The smaller the volatility, the smaller the fluctuation of the estimated value and the better the stability; the larger the volatility, the larger the fluctuation of the estimated value and the worse the stability.

[0107] From the data of the 4 comparison graphs, it can be seen that the smaller the forgetting factor, the shorter the estimation time and the larger the fluctuation of the estimation result; the larger the forgetting factor, the longer the estimation time and the smaller the fluctuation of the estimation result. And the estimation effect of the dynamic forgetting factor is fast and stable.

[0108]

[0109] When the forgetting factor is a fixed value of 0.80 or 0.95, the fluctuation of the estimated value of the bus vehicle mass is relatively large. When the forgetting factor is a fixed value of 0.99 or a dynamically changing value, the fluctuation of the estimated value of the bus vehicle mass is relatively small.

[0110] For FFRLS, when the forgetting factor is 0.99, the volatility is the smallest, which is 35.8 kg / s and 35.7 kg / s; when using the DFFRLS method for estimation, its volatility is less than that of FFRLS with a forgetting factor of 0.99, which is 10 kg / s and 23.8 kg / s.

[0111] When the forgetting factor is 0.80, the estimation time is shorter; when the forgetting factor is 0.99, the estimation volatility is lower, but the forgetting factors are not the same value; while the forgetting factor of DFFRLS is dynamically changing, its estimation time is equivalent to that of FFRLS with a forgetting factor of 0.80, and its estimation volatility is lower than that of FFRLS with a forgetting factor of 0.99.

[0112] In the second aspect, the present invention provides a system that can assist in vehicle mass estimation. The system can be made into a software module and run in the transmission controller. The algorithm can be developed based on Matlab / Simulink. The algorithm module is as Figure 7 shown;

[0113] First, input data such as the running conditions of CCBC buses, the number of passengers, and the gravitational acceleration into the algorithm as Figure 8 shown. The current vehicle speed vVehSpdMeas, driving torque trqDrv, vehicle acceleration aAccVehMeas, and sideslip angle phiSlp can be obtained. By introducing white noise, the simulation of measurement errors of vehicle speed, acceleration, and driving torque is realized. In this example, the noise power of vehicle speed white noise is set to 0.001, and the sampling time is 0.01 s; the noise power of acceleration white noise is set to 0.000001, and the sampling time is 0.01 s; the noise power of driving torque white noise is set to 0.01, and the sampling time is 0.02 s;

[0114] Introduce the current speed measurement and driving torque intoFigure 9 In the algorithm shown, by inputting the air density, air resistance coefficient, frontal area, and tire radius, the algorithm can output the system output quantity E F ;

[0115] Introduce the vehicle acceleration and sideslip angle Figure 10 into the algorithm shown, and by inputting the gravitational acceleration and friction force, the algorithm can output the system observable quantity E a ;

[0116] Input CCBC and K0 Figure 11 into the algorithm shown, and the algorithm can output the dynamic forgetting factor lam.

[0117] Introduce lam and E F , Eα into Figure 12 the algorithm shown, and by introducing the initial mass, maximum mass, and minimum mass of the vehicle, the algorithm can output the vehicle mass Mk at time K.

[0118] Figure 12 The detailed algorithm of the FFRLS algorithm in Figure 13 is shown as follows, where the fcn function is:

[0119]

[0120] The detailed algorithm of the FFRLS_Init algorithm in the FFRLS algorithm is shown as Figure 14 follows, and it is used to record the vehicle values at the previous moment.

[0121] In a third aspect, the present invention provides a device that can measure vehicle data in real time. This device can be mounted on a vehicle to perform real-time vehicle mass estimation.

[0122] There are the following steps, as Figure 15 shown:

[0123] (1) Use a diagnostic instrument to measure the current driving condition of the vehicle. The measured data includes: vehicle speed V, current road surface slope α, and the combined torque Tdr formed by controlling the vehicle throttle and brake pedals.

[0124] (2) By connecting a CAN communication card to the CAN network interface of the vehicle, read the CAN network data on the vehicle in real time.

[0125] (3) Matlab / Simulink uses these real-time read data to estimate the real-time vehicle mass using the above system.

Claims

1. A method for real-time estimation of vehicle mass, characterized in that: The following steps are involved: Constructing the vehicle longitudinal dynamics model: T dr / r tire -(p·A·C w ·v^2) / 2-ε·mg·cos(α)-mg·sin(α)=ma Collect vehicle operation data; determine the parameter r tire Vehicle tire radius, ρ air density, A the frontal area of the vehicle being tested, C W Test environment drag coefficient, ε rolling resistance coefficient of the tested vehicle, g acceleration of gravity, α real-time acquisition slope, v vehicle speed, a acceleration, T dr Drive torque; Construct a vehicle mass calculation method based on the least squares method of the forgetting factor: Where M(k) represents the estimated quality of the k-th step, K(k) represents the gain of the k-th step, and E F (k) represents the output of the k-th step kinetic equation, E α (k) represents the observation at step k, k represents the kth step, and M(k-1) represents the quality of step k-1; According to the vehicle longitudinal dynamics analysis, the system observable quantity E can be obtained: α Model: E α =a+ε·g·cos(·α)+g·sin(·α) According to the vehicle longitudinal dynamics analysis, the system output E can be obtained F Model: Determine the value range and change trend of the dynamic forgetting factor; According to the value range and change trend of the dynamic forgetting factor, the calculation formula of the dynamic forgetting factor is expressed using the exponential increasing formula, and the result is: Where Δt is the iteration period, which is 0.1s; k0 is the number of iterations when the vehicle restarts; Try to find the appropriate values of dynamic factor parameters A, B, and C by exhaustive method; Bring the dynamic forgetting factor into the least squares model of quality to obtain the quality at time k.

2. A vehicle mass real-time estimation method according to claim 1, characterized in that: The vehicle longitudinal dynamics model is converted to: Where r_tire is the tire radius in meters; ρ is the air density, which is 1.29 kg / m^3; A is the frontal area in meters^2; C_W is the drag coefficient; V is the vehicle speed in meters / s; ε is the rolling resistance coefficient; m is the vehicle mass in kilograms; g is the acceleration due to gravity, which is 9.8 N / kg; α is the slope, which is between 0 and 1; and a is the acceleration in meters / s^2.

3. The method for real-time estimation of vehicle mass according to claim 1, characterized in that: The method for setting the dynamic forgetting factor: the value range of the forgetting factor is 0.8-1.

4. The method for real-time estimation of vehicle mass according to claim 1, characterized in that: The least squares model incremental matrix update formula for the quality is: The covariance matrix update formula is: P(k-1) represents the covariance at the k-1th step, P(k) represents the covariance at the kth step, and λ represents the forgetting factor.

5. A system for assisting vehicle mass estimation, characterized in that: The system includes the following modules: Vehcle module that calculates vehicle speed, acceleration, torque, and slope; Mass_RLS module for estimating vehicle mass; The Mass_RLS module adopts any one of the vehicle mass real-time estimation methods of claims 1-4, and the Mass_RLS module includes: lam module for calculating dynamic forgetting factor; EF_Cal module for calculating system observations; EA_Cal module that calculates system output; FFRLS module of recursive least squares method; The FFRLS module includes: FFRLS_Init module used to record K-1 time parameters.

6. A device for real-time mass estimation on a vehicle, characterized in that: It includes the following modules: data acquisition module, a device for collecting vehicle operation data; Data transmission module, which transmits the collected data to the CAN network in the computer software Matlab; The mass calculation module receives data from the CAN network and transmits it to Matlab, and uses the system for assisting vehicle mass estimation as described in claim 5 to perform mass estimation.

Images