Commercial vehicle quality and road slope decoupling estimation method for eliminating unstable working conditions

Abstract

The invention belongs to the technical field of automobiles, and particularly relates to a commercial vehicle quality and road slope decoupling estimation method for eliminating unstable working conditions. Comprising the following steps: 1, analyzing and estimating working conditions through a fuzzy controller, synthesizing a confidence factor, eliminating the confidence factor when the confidence factor is smaller than a threshold value under an unstable working condition, suspending quality estimation, and only triggering quality estimation under a stable working condition that the confidence factor is greater than or equal to the threshold value; 2, estimating the road gradient based on a Kalman filtering algorithm, and completing decoupling of the road gradient and the commercial vehicle mass; and step 3, under a stable working condition that the confidence factor is greater than or equal to a threshold value, completing online quality estimation of the decoupled commercial vehicle based on a recursive least square method with a forgetting factor. According to the method, the commercial vehicle mass and the road gradient can be accurately estimated, and a basic basis is provided for commercial vehicle chassis integrated control.

Description

Technical Field

[0001] The invention belongs to the field of automobile technology, and in particular is a decoupling estimation method for commercial vehicle mass and road slope excluding unstable working conditions. Background Art

[0002] Vehicle chassis integrated control is an important part of vehicle intelligent technology, which has a profound impact on vehicle safety, economy, comfort and other performance. Vehicle mass and road slope are key parameters in vehicle chassis integrated control, which are often difficult to measure directly through sensors and are mostly obtained through online estimation. Therefore, how to accurately estimate the mass and road slope of commercial vehicles is of great significance.

[0003] At present, vehicle mass estimation is usually based on vehicle longitudinal dynamics modeling and estimated using a recursive least squares algorithm. However, most simplified vehicle dynamics models used for vehicle mass estimation do not consider the large impact of commercial vehicle rotational mass, which will lead to a decrease in estimation accuracy. In practical applications, it is necessary to estimate the vehicle mass and road slope online. The estimation method based on vehicle dynamics has attracted much attention due to its simple signal acquisition. However, in the vehicle longitudinal dynamics model, there is a coupling relationship between vehicle mass and road slope, and online estimation is difficult. Some scholars use the least squares or Kalman filter algorithm to jointly estimate vehicle mass and road slope, but this method can only have a good estimation effect when the vehicle is running smoothly at low speed. Under unstable conditions such as high speed, large acceleration, braking, and gear shifting, the estimation effect will fluctuate greatly. The poor applicability of the estimation method needs to be improved urgently. Summary of the invention

[0004] To solve the above problems, the present invention provides a decoupling estimation method for commercial vehicle mass and road slope that eliminates unstable working conditions, which can accurately estimate the commercial vehicle mass and road slope, and provide a basic basis for the integrated control of commercial vehicle chassis.

[0005] The technical solution of the present invention is described as follows in conjunction with the accompanying drawings:

[0006] A method for estimating the mass of a commercial vehicle and the road slope by decoupling the vehicle mass and the road slope without unstable working conditions comprises the following steps:

[0007] Step 1: Analyze and estimate the working conditions through the fuzzy controller and synthesize the confidence factor. Under unstable working conditions, if the confidence factor is less than the threshold value, it will be excluded and the quality estimation will be suspended. The quality estimation will be triggered only under stable working conditions where the confidence factor is greater than or equal to the threshold value.

[0008] Step 2: Estimate the road slope based on the Kalman filter algorithm to decouple the road slope from the mass of the commercial vehicle;

[0009] Step 3: Under stable working conditions where the confidence factor is greater than or equal to the threshold value, the decoupled commercial vehicle mass is estimated online based on the recursive least squares method with a forgetting factor.

[0010] Furthermore, the specific method of step one is as follows:

[0011] 11) Design fuzzy controller;

[0012] 111) Select the engine effective torque T e , longitudinal velocity v, absolute value of longitudinal acceleration |dv / dt|, maximum value of each brake chamber pressure P i,max As fuzzy controller input;

[0013] 112) According to the simulation experiment, the domain and feature set range of each input variable are determined, and the fuzzy subsets are defined as small, medium, and large;

[0014] 113) Selecting a bilateral membership function and a Gaussian membership function as the membership function;

[0015] 114) Define the output variable as Y, the domain range as (0,1), and the fuzzy subsets as poor, medium, and good;

[0016] 115) Establish Mandani type fuzzy rule: In the input variable engine effective torque T e , longitudinal velocity v, absolute value of longitudinal acceleration |dv / dt|, maximum value of each brake chamber pressure P i,max And the output variable Y, a total of 15 fuzzy subsets are set with fuzzy rules:

[0017] Generate a fuzzy reasoning surface according to the fuzzy rules, that is, obtain the output value Y through the fuzzy controller;

[0018] 12) The state of the vehicle in gear, i.e. the clutch is fully engaged, is taken as 1, and the state of neutral gear, i.e. the clutch is not fully engaged, is taken as 0; a confidence factor K is set as a criterion for judging whether the vehicle is in a stable state suitable for vehicle mass estimation; when the vehicle is in gear, the confidence factor K is equal to the output value Y of the fuzzy controller; when the vehicle is in neutral gear, the confidence factor K is taken as 0; a threshold value is set to 0.6, and when the confidence factor K ≥ 0.6, it indicates that the vehicle is in a stable condition and is suitable for vehicle mass estimation; when the confidence factor K < 0.6, it indicates that the vehicle is in an unstable condition that is not suitable for mass estimation, and the condition is suspended, and the estimation is continued when the stable condition is restored.

[0019] Furthermore, the fuzzy rules are as follows:

[0020] (1) For the four input parameters T tq , P i,max , v xand |dv / dt|, when any one of them is larger, the output variable Y is considered to be worse;

[0021] (2) For v x and |dv / dt| is small, and T tq , P i,max For small or medium working conditions, it is better to judge the output variable Y;

[0022] (3) For v x and |dv / dt| is small or medium, and T tq , P i,max For smaller working conditions, it is better to judge the output variable Y;

[0023] (4) For v x and |dv / dt|, T tq Small or medium, and P i,max For medium working conditions, the output variable Y is judged to be medium;

[0024] (5) For v x and P i,max , T tq When |dv / dt| is small or medium and the working condition is medium, the output variable Y is judged to be medium.

[0025] Furthermore, the specific method of step 2 is as follows:

[0026] 21) Ignoring the pitch, yaw and roll characteristics of the vehicle, the following vehicle kinematic model is established:

[0027] a x =dv / dt+gi (1)

[0028] In the formula, a x is the vehicle acceleration sensor signal; v is the vehicle longitudinal velocity; dv / dt is the vehicle longitudinal acceleration; i is the road slope; g is the gravity acceleration; where the acceleration sensor signal a x Contains vehicle speed information and slope information;

[0029] 22) The time differential of the slope di / dt = 0; the vehicle kinematic model (1) is deformed to obtain the following differential equation:

[0030]

[0031] 23) The vehicle kinematic model (2) in differential form is converted into the form of state space equations and discretized as follows:

[0032]

[0033] In the formula, the state variable State Space Matrix State Space Matrix Enter u k =a x,k ; Observed value z k =v k ; Observation matrix H = [1 0]; where Δt is the unit time step; v k is the speed measured at time k; i k is the estimated slope at time k; a x,k is the acceleration signal at time k; W k is the process noise matrix at time k, which represents the deviation between the system state space equation and the actual process; V k is the measurement noise matrix at time k, representing the sensor accuracy error;

[0034] 24) Assume that the process noise W of the discrete system k and measurement noise V k is an uncorrelated white noise with a mean of zero and variances of Q and R respectively, then the prior estimation part of the Kalman filter is expressed as:

[0035]

[0036] In the formula, is the prior state estimate at time k; are the prior estimates of speed and road slope at time k respectively;

[0037] 25) The prior covariance matrix is ​​calculated as:

[0038]

[0039] In the formula, is the prior error covariance matrix at time k; P k-1 is the error covariance matrix after updating at time k-1; Q is the process noise covariance matrix;

[0040] 26) The Kalman gain is calculated as:

[0041]

[0042] In the formula, K k is the Kalman gain at time k; is the a priori error covariance matrix at time k; H = [1 0] is the observation matrix of the system; R is the measurement noise covariance matrix;

[0043] 27) The posterior estimate is calculated as:

[0044]

[0045] In the formula, is the posterior estimate at time k; is the prior state estimate at time k; K k is the Kalman gain at time k; z k =v k is the measured value; are the posterior and prior estimates of the speed and road slope at time k, respectively, v k is the speed measurement value at time k;

[0046] 28) Error covariance matrix update:

[0047]

[0048] Where P k is the error covariance matrix after updating at time k; E is the unit matrix; K k is the Kalman gain at time k; H is the observation matrix of the system; is the prior error covariance matrix at time k;

[0049] Among them, the posterior estimate of the road slope at time k calculated in the posterior estimate (7) is This is the estimated road slope i.

[0050] Furthermore, the specific method of step three is as follows:

[0051] 31) Establish a vehicle dynamics model; the vehicle longitudinal dynamics model used only considers the longitudinal force of the vehicle when driving in a straight line, and the longitudinal force balance equation of the vehicle can be obtained as follows:

[0052] F x =F w +F i +F j +F f (9)

[0053] Where, the longitudinal force F x Represents driving force F t or braking force F b Driving force Braking force Air resistance Slope resistance F i = mgi; acceleration resistance Rolling resistance F f = mgf; T tq is the effective torque of the engine; i0 is the main reduction transmission ratio; i g is the transmission ratio; η t is the total transmission efficiency; r r is the wheel rolling radius; P f is the front axle brake pressure; Pr is the brake pressure of the rear axle brake; K bf is the braking efficiency factor of the front axle brake; K br is the braking efficiency factor of the rear axle brake; C D is the air resistance coefficient of the commercial vehicle; A is the frontal area of ​​the commercial vehicle; v is the longitudinal speed; m is the vehicle mass; g is the acceleration of gravity; i is the tangent value of the slope angle α, i.e. the slope; δ is the rotation mass conversion coefficient; is the vehicle acceleration; f is the rolling resistance coefficient of the tire;

[0054] 32) The vehicle mass m and the road slope i have a slope resistance F. i =mgi, there is coupling. Substituting the road slope i obtained in the a posteriori estimation (7) into the vehicle longitudinal force balance equation (9) yields the vehicle mass m identification expression:

[0055]

[0056] Where, T tq is the effective torque of the engine; i0 is the main reduction transmission ratio; i g is the transmission ratio; η t is the total transmission efficiency; r r is the wheel rolling radius; P f is the front axle brake pressure; P r is the brake pressure of the rear axle brake; K bf is the braking efficiency factor of the front axle brake; K br is the braking efficiency factor of the rear axle brake; C D is the air resistance coefficient of the commercial vehicle; A is the frontal area of ​​the commercial vehicle; v is the longitudinal velocity; m is the vehicle mass; g is the acceleration of gravity; i is the tangent value of the slope angle α; δ is the rotation mass conversion coefficient; is the vehicle acceleration, f is the rolling resistance coefficient of the tire;

[0057] 33) The vehicle mass m is estimated by the least squares method, and the vehicle mass m identification expression (10) is converted into the least squares format:

[0058]

[0059] In the formula, is the estimated value of vehicle mass at time k; is the k-time input; Output at time k;

[0060] 34) Improve the least squares method (11) into the recursive least squares method:

[0061]

[0062] In the formula, K is the estimated value of the vehicle mass at time k; k is the gain at time k; P k is the gain covariance at time k;

[0063] 35) Substitute the forgetting factor μ into the recursive least squares method (12) to form the recursive least squares method with forgetting factor:

[0064]

[0065] 36) Set the mass m and Kalman gain K in the recursive least squares method with forgetting factor (13) k And the covariance matrix P k The initial value of , that is, the vehicle mass can be estimated.

[0066] Furthermore, the forgetting factor is set to 0.95 within 0-3s; and the forgetting factor value is gradually increased starting from the 3rd second of the estimation time.

[0067] The beneficial effects of the present invention are:

[0068] 1) The vehicle dynamics model constructed by the present invention adds a rotational mass conversion coefficient, which well characterizes the rotational mass of commercial vehicles and weakens the influence of rotational mass on vehicle mass estimation

[0069] 2) The estimation algorithm designed based on the Kalman filter algorithm of the present invention can independently estimate the road slope without the need for vehicle mass, so that the vehicle mass and road slope can be decoupled

[0070] 3) The present invention estimates the vehicle mass based on the recursive least squares method with a forgetting factor, which improves the ability of observation data to update the estimation results and ensures the accuracy of mass estimation.

[0071] 4) The present invention designs a working condition discrimination module based on a fuzzy controller, and stops estimating when it is identified through a confidence factor that the vehicle is in an unstable working condition, thereby eliminating the interference of unstable working conditions on the estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0073] Figure 1A structural diagram of a method for decoupling the mass of a commercial vehicle and the road slope to eliminate unstable working conditions according to the present invention;

[0074] Figure 2 It is a schematic diagram of vehicle kinematic information;

[0075] Figure 3 Schematic diagram of the input variable membership function;

[0076] Figure 4 It is a schematic diagram of the output variable membership function;

[0077] Figure 5 It is a schematic diagram of the vehicle longitudinal dynamics model;

[0078] Figure 6 Schematic diagram of the forgetting factor changing over time;

[0079] Figure 7 This is a schematic diagram of the road slope estimation result;

[0080] Figure 8 Schematic diagram of commercial vehicle mass estimation results. DETAILED DESCRIPTION

[0081] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0082] Embodiment 1

[0083] See also Figure 1 The present invention provides a method for decoupling the mass and road slope of a commercial vehicle excluding unstable working conditions. First, the signals of various vehicle sensors are input into a working condition discrimination module to identify and exclude unstable working conditions such as high speed, large acceleration, braking, and gear shifting that are not suitable for vehicle mass estimation; then, a road slope estimation model is established based on the longitudinal kinematics of the vehicle, and the road slope is estimated using a Kalman filter algorithm; finally, the recursive least squares method is used to estimate the vehicle mass under stable working conditions to achieve decoupling estimation of road slope and vehicle mass, as follows:

[0084] Step 1: Analyze and estimate the working conditions through the fuzzy controller and synthesize the confidence factor. Under unstable working conditions, if the confidence factor is less than the threshold value, it will be excluded and the quality estimation will be suspended. The quality estimation will be triggered only under stable working conditions where the confidence factor is greater than or equal to the threshold value. The details are as follows:

[0085] 11) Design fuzzy controller;

[0086] 111) Select the engine effective torque Te , longitudinal velocity v, absolute value of longitudinal acceleration |dv / dt|, maximum value of each brake chamber pressure P i,max As fuzzy controller input;

[0087] 112) According to the simulation experiment, the domain and feature set range of the input variables are determined, and the fuzzy subsets are defined as small, medium, and large, as shown in Table 1 below:

[0088] Table 1 Fuzzy domain of input variables

[0089]

[0090] 113) In order to better characterize the membership change trend of each variable, the bilateral membership function and Gaussian membership function are selected as the membership function, such as Figure 3 shown.

[0091] 114) Define the output variable as Y, the domain range is (0,1), and the fuzzy subsets are defined as poor, medium, and good. The corresponding output ranges are poor (0-0.5), medium (0.5-0.7), and good (0.7-1.0), such as Figure 4 shown.

[0092] 115) Establish Mandani type fuzzy rule: In the input variable engine effective torque T e , longitudinal velocity v, absolute value of longitudinal acceleration |dv / dt|, maximum value of each brake chamber pressure P i,max The following five fuzzy rules are set among the 15 fuzzy subsets:

[0093] (1) For the four input parameters T tq , P i,max , v x and |dv / dt|, when any one of them is larger, the output variable Y is considered to be worse;

[0094] (2) For v x and |dv / dt| is small, and T tq , P i,max For small or medium working conditions, it is better to judge the output variable Y;

[0095] (3) For v x and |dv / dt| is small or medium, and T tq , P i,max For smaller working conditions, it is better to judge the output variable Y;

[0096] (4) For v x and |dv / dt|, T tq Small or medium, and P i,maxFor medium working conditions, the output variable Y is judged to be medium;

[0097] (5) For v x and P i,max , T tq When |dv / dt| is small or medium and the working condition is medium, the output variable Y is judged to be medium.

[0098] The fuzzy reasoning surface is generated according to the above fuzzy rules, and the output value Y can be obtained through the fuzzy controller.

[0099] 12) Since the gear shifting process will cause the vehicle's power transmission to be interrupted, which will interfere with the vehicle mass estimation, the vehicle in gear, that is, the clutch is fully engaged, is set to 1, and the neutral gear, that is, the clutch is not fully engaged, is set to 0; the confidence factor K is set as the criterion for judging whether the vehicle is in a stable state suitable for vehicle mass estimation. When the vehicle is in gear, the confidence factor K is equal to the fuzzy controller output value Y; when the vehicle is neutral, the confidence factor K is 0; the threshold value is set to 0.6, and when the confidence factor K ≥ 0.6, it indicates that the vehicle is in a stable condition and is suitable for vehicle mass estimation; when the confidence factor K < 0.6, it indicates that the vehicle is in an unstable condition that is not suitable for mass estimation, and it is paused, and the estimation is continued when the stable condition is restored.

[0100] Step 2: See Figure 2 , the road slope is estimated based on the Kalman filter algorithm, and the road slope is decoupled from the commercial vehicle mass, as follows:

[0101] 21) The road slope is mainly based on the vehicle acceleration sensor signal a x The vehicle longitudinal velocity signal v is estimated online. Ignoring the pitch, yaw and roll characteristics of the vehicle, the following vehicle kinematic model is established:

[0102] a x =dv / dt+gi (1)

[0103] In the formula, a x is the vehicle acceleration sensor signal; v is the vehicle longitudinal velocity; dv / dt is the vehicle longitudinal acceleration; i is the road slope; g is the gravity acceleration; where the acceleration sensor signal a x Contains vehicle speed information and slope information;

[0104] 22) Since the road slope changes slowly over time during the actual driving process of the vehicle, it can be approximately considered that the change in slope in a very short period of time is zero, that is, the time differential of slope di / dt = 0. Then the vehicle kinematic model (1) can be transformed to obtain the following differential equation:

[0105]

[0106] In the formula, a x is the vehicle acceleration sensor signal; v is the vehicle longitudinal velocity; dv / dt is the vehicle longitudinal acceleration; i is the road slope; g is the gravitational acceleration;

[0107] 23) In order to minimize the mean square error of the slope estimation value, the present invention adopts the Kalman filter algorithm to perform online estimation of the road slope. The present invention converts the vehicle kinematic model (2) in differential form into the form of state space equation and discretizes it to facilitate actual calculation, as shown below:

[0108]

[0109] In the formula, the state variable State Space Matrix State Space Matrix Enter u k =a x,k ; Observed value z k =v k ; Observation matrix H = [1 0]; where g is the acceleration due to gravity; Δt is the unit time step; v k is the speed measured at time k; i k is the estimated slope at time k; a x,k is the acceleration signal at time k; W k is the process noise matrix at time k, which represents the deviation between the system state space equation and the actual process; V k is the measurement noise matrix at time k, representing the sensor accuracy error;

[0110] 24) Assume that the process noise W of the discrete system k and measurement noise V k is an uncorrelated white noise with a mean of zero and variances of Q and R respectively, then the prior estimation part of the Kalman filter is expressed as:

[0111]

[0112] In the formula, is the prior state estimate at time k; is the state space matrix; is the state space matrix; u k =a x,k ; are the prior estimates of speed and road slope at time k; g is the acceleration of gravity; Δt is the unit time step; a x,k is the acceleration signal at time k;

[0113] 25) The prior covariance matrix is ​​calculated as:

[0114]

[0115] In the formula, is the prior error covariance matrix at time k; P k-1 is the error covariance matrix after updating at time k-1; Q is the process noise covariance matrix; g is the gravitational acceleration; Δt is the unit time step;

[0116] 26) The Kalman gain is calculated as:

[0117]

[0118] In the formula, K k is the Kalman gain at time k; is the a priori error covariance matrix at time k; H = [1 0] is the observation matrix of the system; R is the measurement noise covariance matrix;

[0119] 27) The posterior estimate is calculated as:

[0120]

[0121] In the formula, is the posterior estimate at time k; is the prior state estimate at time k; K k is the Kalman gain at time k; z k =v k is the measured value; H = [1 0] is the observation matrix of the system; where are the posterior and prior estimates of the speed and road slope at time k, respectively, v k is the speed measurement value at time k;

[0122] 28) Error covariance matrix update:

[0123]

[0124] Where P k is the error covariance matrix after updating at time k; E is the unit matrix; K k is the Kalman gain at time k; H is the observation matrix of the system; P k - is the prior error covariance matrix at time k;

[0125] The posterior estimate of the road slope at time k calculated in the posterior estimate (7) is This is the estimated road slope i.

[0126] Step 3: Online estimation of commercial vehicle mass based on recursive least squares method, as follows:

[0127] 31) In order to achieve accurate commercial vehicle mass estimation, it is necessary to first establish a suitable vehicle dynamics model. Figure 5 As shown, a vehicle dynamics model is established; the vehicle longitudinal dynamics model used only considers the longitudinal force on the vehicle when it is driving in a straight line, and the longitudinal force balance equation of the vehicle can be obtained as follows:

[0128] F x =F w +F i +F j +F f (9)

[0129] Where, the longitudinal force F x Represents driving force F t or braking force F b Driving force Braking force Air resistance Slope resistance F i = mgi; acceleration resistance Rolling resistance F f = mgf; T tq is the effective torque of the engine; i0 is the main reduction transmission ratio; i g is the transmission ratio; η t is the total transmission efficiency; r r is the wheel rolling radius; P f is the front axle brake pressure; P r is the brake pressure of the rear axle brake; K bf is the braking efficiency factor of the front axle brake; K br is the braking efficiency factor of the rear axle brake; C D is the air resistance coefficient of the commercial vehicle; A is the frontal area of ​​the commercial vehicle; v is the longitudinal speed; m is the vehicle mass; g is the acceleration of gravity; i is the tangent value of the slope angle α, i.e. the slope; δ is the rotation mass conversion coefficient; is the vehicle acceleration; f is the rolling resistance coefficient of the tire;

[0130] 32) From the vehicle longitudinal force balance equation (9), we know that the vehicle mass m and the road slope i have a slope resistance F i =mgi, there is coupling. Substituting the road slope i obtained in the a posteriori estimation (7) into the vehicle longitudinal force balance equation (9) yields the vehicle mass m identification expression:

[0131]

[0132] Where, T tq is the effective torque of the engine; i0 is the main reduction transmission ratio; i g is the transmission ratio; η t is the total transmission efficiency; rr is the wheel rolling radius; P f is the front axle brake pressure; P r is the brake pressure of the rear axle brake; K bf is the braking efficiency factor of the front axle brake; K rf is the braking efficiency factor of the rear axle brake; C D is the air resistance coefficient of the commercial vehicle; A is the frontal area of ​​the commercial vehicle; v is the longitudinal velocity; m is the vehicle mass; g is the acceleration of gravity; i is the tangent value of the slope angle α; δ is the rotation mass conversion coefficient; is the vehicle acceleration, f is the rolling resistance coefficient of the tire;

[0133] 33) Since the least squares method has a simple principle, is easy to calculate, and is widely used, the present invention uses the least squares method to estimate the vehicle mass m and converts the vehicle mass m identification expression (10) into the least squares format:

[0134]

[0135] In the formula, is the estimated value of vehicle mass at time k; is the k-time input; Output at time k;

[0136] 34) Since the vehicle mass needs to be estimated in real time, the least squares method cannot meet the real-time requirements. The least squares method (11) is improved to the recursive least squares method:

[0137]

[0138] In the formula, K is the estimated value of the vehicle mass at time k; k is the gain at time k; P k is the gain covariance at time k; is the k-time input; Output at time k;

[0139] 35) However, after multiple iterations of the recursive least squares method, the gain K k and the covariance P k It will gradually decrease, eventually resulting in data saturation and gradually losing the ability of real-time estimation. The input of new observation data is difficult to correct the old estimation results with errors. To address this defect, the present invention introduces a forgetting factor μ that changes with the estimation time to correct the recursive least squares method, so as to accelerate the convergence of the estimation results and weaken the fluctuation of the estimation results.

[0140] Substitute the forgetting factor μ into the recursive least squares method (12) to form the recursive least squares method with forgetting factor:

[0141]

[0142] In the formula, K is the estimated value of vehicle mass at time k; k is the gain at time k; P k is the gain covariance at time k; is the k-time input; Output at time k;

[0143] 36) Set the mass m and Kalman gain K in the recursive least squares method with forgetting factor (13) k And the covariance matrix P k The initial value of , that is, the vehicle mass can be estimated.

[0144] See also Figure 6 In order to improve the iterative update capability of the recursive least squares method at the beginning of the estimation stage, the forgetting factor is set to 0.95 within 0-3s; after a period of estimation, the vehicle mass obtained is close to the true value. In order to reduce the fluctuation of subsequent estimation results, the forgetting factor value is gradually increased from the 3rd second of the estimation time to appropriately weaken the mass fluctuation caused by the update of the recursive least squares method.

[0145] In summary, the present invention can accurately estimate the mass of a commercial vehicle and the road slope, and provides a basic basis for the integrated control of the commercial vehicle chassis.

[0146] Embodiment 2

[0147] The present invention performs simulation testing on a commercial vehicle mass and road slope decoupling estimation method designed to eliminate unstable working conditions based on a MATLAB / Simulink-CarSim joint simulation platform.

[0148] Road slope estimation simulation test Figure 7 As shown in the figure, the simulated road is flat road → 0.1 slope road → flat road → -0.1 slope road → flat road. It can be seen that when the vehicle encounters a sudden road slope, the slope estimation result can quickly converge to the true value within 2 seconds, and the estimation effect is good.

[0149] Commercial vehicle mass estimation part Figure 8 As shown in the figure, the actual mass of the commercial vehicle is set to 4505kg, the initial mass is 4000kg, and the vehicle starts to accelerate from the 0th second. It can be seen that the mass estimation curve has three sections with a constant slope of zero after the start. Due to excessive vehicle acceleration or gear shifting, the built-in confidence factors in these intervals are less than the threshold value. These working conditions are considered to be unstable working conditions that are not suitable for mass estimation and are excluded. This effectively improves the accuracy of commercial vehicle mass estimation and can provide more accurate vehicle mass parameters for vehicle motion control.

[0150] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A decoupling estimation method for commercial vehicle mass and road slope excluding unstable working conditions, characterized in that: The following steps are involved: Step 1: Analyze and estimate the working conditions through the fuzzy controller and synthesize the confidence factor. Under unstable working conditions, if the confidence factor is less than the threshold value, it will be excluded and the quality estimation will be suspended. The quality estimation will be triggered only under stable working conditions where the confidence factor is greater than or equal to the threshold value. Step 2: Estimate the road slope based on the Kalman filter algorithm to decouple the road slope from the mass of the commercial vehicle; Step 3: Under stable working conditions where the confidence factor is greater than or equal to the threshold value, the decoupled commercial vehicle mass is estimated online based on the recursive least squares method with a forgetting factor.

2. The method for decoupling the mass of a commercial vehicle and the road slope to eliminate unstable working conditions according to claim 1 is characterized in that: The specific method of step one is as follows: 11) Design fuzzy controller; 111) Select the engine effective torque T e , longitudinal velocity v, absolute value of longitudinal acceleration |dv / dt|, maximum value of each brake chamber pressure P i,max As fuzzy controller input; 112) According to the simulation experiment, the domain and feature set range of each input variable are determined, and the fuzzy subsets are defined as small, medium, and large; 113) Selecting a bilateral membership function and a Gaussian membership function as the membership function; 114) Define the output variable as Y, the domain range as (0,1), and the fuzzy subsets as poor, medium, and good; 115) Establish Mandani type fuzzy rule: In the input variable engine effective torque T e , longitudinal velocity v, absolute value of longitudinal acceleration |dv / dt|, maximum value of each brake chamber pressure P i,max And the output variable Y, a total of 15 fuzzy subsets are set with fuzzy rules: Generate a fuzzy reasoning surface according to the fuzzy rules, that is, obtain the output value Y through the fuzzy controller; 12) The state of the vehicle in gear, i.e. the clutch is fully engaged, is taken as 1, and the state of neutral gear, i.e. the clutch is not fully engaged, is taken as 0; a confidence factor K is set as a criterion for judging whether the vehicle is in a stable state suitable for vehicle mass estimation; when the vehicle is in gear, the confidence factor K is equal to the output value Y of the fuzzy controller; when the vehicle is in neutral gear, the confidence factor K is taken as 0; a threshold value is set to 0.6, and when the confidence factor K ≥ 0.6, it indicates that the vehicle is in a stable condition and is suitable for vehicle mass estimation; when the confidence factor K < 0.6, it indicates that the vehicle is in an unstable condition that is not suitable for mass estimation, and the condition is suspended, and the estimation is continued when the stable condition is restored.

3. The method for decoupling the commercial vehicle mass and the road slope to estimate the unstable working conditions according to claim 2, characterized in that: The fuzzy rules are as follows: (1) For the four input parameters T tq , P i,max , v x and |dv / dt|, when any one of them is larger, the output variable Y is considered to be worse; (2) For v x and |dv / dt| is small, and T tq , P i,max For small or medium working conditions, it is better to judge the output variable Y; (3) For v x and |dv / dt| is small or medium, and T tq , P i,max For smaller working conditions, it is better to judge the output variable Y; (4) For v x and |dv / dt|, T tq Small or medium, and P i,max For medium working conditions, the output variable Y is judged to be medium; (5) For v x and P i,max , T tq When |dv / dt| is small or medium and the working condition is medium, the output variable Y is judged to be medium.

4. The method for decoupling the commercial vehicle mass and the road slope to estimate the unstable working condition according to claim 1, characterized in that: The specific method of step 2 is as follows: 21) Ignoring the pitch, yaw and roll characteristics of the vehicle, the following vehicle kinematic model is established: a x =dv / dt+gi (1) In the formula, a x is the vehicle acceleration sensor signal; v is the vehicle longitudinal velocity; dv / dt is the vehicle longitudinal acceleration; i is the road slope; g is the gravity acceleration; where the acceleration sensor signal a x Contains vehicle speed information and slope information; 22) The time differential of the slope di / dt = 0; the vehicle kinematic model (1) is deformed to obtain the following differential equation: 23) The vehicle kinematic model (2) in differential form is converted into the form of state space equations and discretized as follows: In the formula, the state variable State Space Matrix State Space Matrix Enter u k =a x,k ; Observed value z k =v k ; Observation matrix H = [1 0]; where Δt is the unit time step; v k is the speed measured at time k; i k is the estimated slope at time k; a x,k is the acceleration signal at time k; W k is the process noise matrix at time k, which represents the deviation between the system state space equation and the actual process; V k is the measurement noise matrix at time k, representing the sensor accuracy error; 24) Assume that the process noise W of the discrete system k and measurement noise V k is an uncorrelated white noise with a mean of zero and variances of Q and R respectively, then the prior estimation part of the Kalman filter is expressed as: In the formula, is the prior state estimate at time k; are the prior estimates of speed and road slope at time k respectively; 25) The prior covariance matrix is ​​calculated as: In the formula, is the prior error covariance matrix at time k; P k-1 is the error covariance matrix after updating at time k-1; Q is the process noise covariance matrix; 26) The Kalman gain is calculated as: In the formula, K k is the Kalman gain at time k; R is the measurement noise covariance matrix; 27) The posterior estimate is calculated as: In the formula, is the posterior estimate at time k; is the prior state estimate at time k; K k is the Kalman gain at time k; z k =v k is the measured value; are the posterior and prior estimates of the speed and road slope at time k, respectively, v k is the speed measurement value at time k; 28) Error covariance matrix update: Where P k is the error covariance matrix after updating at time k; E is the unit matrix; K k is the Kalman gain at time k; H is the observation matrix of the system; is the prior error covariance matrix at time k; The posterior estimate of the road slope at time k calculated in the posterior estimate (7) is This is the estimated road slope i.

5. The method for decoupling the commercial vehicle mass and the road slope to estimate the unstable working condition according to claim 4, characterized in that: The specific method of step three is as follows: 31) Establish a vehicle dynamics model; the vehicle longitudinal dynamics model used only considers the longitudinal force of the vehicle when driving in a straight line, and the longitudinal force balance equation of the vehicle can be obtained as follows: F x =F w +F i +F j +F f (9) Where, the longitudinal force F x Represents driving force F t or braking force F b Driving force Braking force Air resistance Slope resistance F i = mgi; acceleration resistance Rolling resistance F f = mgf; T tq is the effective torque of the engine; i0 is the main reduction transmission ratio; i g is the transmission ratio; η t is the total transmission efficiency; r r is the wheel rolling radius; P f is the front axle brake pressure; P r is the brake pressure of the rear axle brake; K bf is the braking efficiency factor of the front axle brake; K br is the braking efficiency factor of the rear axle brake; C D is the air resistance coefficient of the commercial vehicle; A is the frontal area of ​​the commercial vehicle; v is the longitudinal speed; i is the tangent value of the slope angle α, that is, the slope; δ is the rotation mass conversion coefficient; is the vehicle acceleration; f is the rolling resistance coefficient of the tire; 32) The vehicle mass m and the road slope i have a slope resistance F. i =mgi, there is coupling. Substituting the road slope i obtained in the a posteriori estimation (7) into the vehicle longitudinal force balance equation (9) yields the vehicle mass m identification expression: 33) The vehicle mass m is estimated by the least squares method, and the vehicle mass m identification expression (10) is converted into the least squares format: In the formula, is the estimated value of vehicle mass at time k; is the k-time input; Output at time k; 34) Improve the least squares method (11) into the recursive least squares method: In the formula, K is the estimated value of vehicle mass at time k; k is the gain at time k; P k is the gain covariance at time k; 35) Substitute the forgetting factor μ into the recursive least squares method (12) to form the recursive least squares method with forgetting factor: 36) Set the mass m and Kalman gain K in the recursive least squares method with forgetting factor (13) k And the covariance matrix P k The initial value of , that is, the vehicle mass can be estimated.

6. A method for estimating the mass of a commercial vehicle and the road slope by decoupling the vehicle mass and the road slope without unstable working conditions according to claim 5, characterized in that: The forgetting factor is set to 0.95 within 0-3s; the forgetting factor value is gradually increased starting from the 3rd second of the estimation time.

Images