A vehicle mass estimation method based on engine torque estimation

Abstract

The present invention discloses a vehicle mass estimation method based on engine torque estimation, aiming to solve the shortcomings of existing vehicle mass estimation methods such as low estimation accuracy, difficult to obtain input signals, and poor robustness. By obtaining information such as the longitudinal speed, longitudinal acceleration, wheel speed, intake mass flow, air-fuel ratio, etc. of the vehicle from the vehicle network bus and sensors, the engine torque and real-time gear information are first calculated to create a more ideal working condition for the mass estimation algorithm. The vehicle mass estimation working condition is judged using the established fuzzy rules, and then the current vehicle mass is updated using the recursive least squares method. The present invention mainly uses basic vehicle bus signals and sensor information, has the advantages of wide application range, high accuracy and strong robustness, and provides a reliable vehicle mass signal input for the vehicle control system.

Description

Technical Field

[0001] The present invention relates to the field of vehicle parameter estimation, and more specifically, to a vehicle mass estimation method based on engine torque estimation. Background Art

[0002] With the rapid development of vehicle electronic control systems in recent years, the accuracy of parameter estimation has an increasingly significant impact on vehicle control effects. Among them, the mass of the vehicle is closely related to the vehicle's dynamic properties, and the mass of the vehicle changes in real time during the entire driving process. In particular, the weight of heavy commercial vehicles varies greatly when fully loaded and unloaded. Large mass changes will have a great impact on the basic performance of the vehicle, such as braking and handling stability. Achieving accurate online estimation of the vehicle's mass is one of the necessary conditions for realizing automotive electronic control technology.

[0003] Real-time estimation of vehicle mass can use a dynamic vehicle weighing system (WIM), which consists of a series of sensors and software that can calculate the current vehicle's wheel weight, axle weight, and total weight data. However, this method is relatively expensive, so it is rarely used in actual vehicles. Common vehicle real-time mass estimation methods are usually based on the longitudinal dynamic balance equation of the vehicle, using least squares, Kalman filtering, or other derivative methods to estimate the vehicle mass. These dynamic methods require obtaining time-varying information such as engine torque and transmission ratio from the vehicle bus, but not all bus data specified by the manufacturer can provide relevant information. Therefore, the applicability of related methods still needs to be further improved.

[0004] Among the existing patents, for example, the Chinese patent authorization announcement number is CN 109477754 B, the authorization announcement date is January 8, 2021, and the invention name is "Method and device for determining the estimated value of the total mass of a motor vehicle". The invention uses communication network data and vehicle parameters to estimate the vehicle parameters at the next moment based on parameters such as the vehicle mass at the previous moment, but this method does not consider the influence of traction acquisition method and accuracy; for example, the Chinese patent publication number is CN 113978473 A, the publication date is January 28, 2022, and the invention name is "A vehicle mass and road slope estimation method". The invention combines the vehicle longitudinal dynamics model and the longitudinal kinematics model, and uses a neural network algorithm to jointly estimate the mass and slope of the current vehicle, but the accuracy of the result will be affected by the neural network training samples, and the adaptability and robustness are poor.

[0005] In summary, the current online vehicle mass estimation methods rarely consider the acquisition methods of partial data such as torque and transmission ratio, and have poor adaptability. Therefore, it is necessary to provide such a vehicle mass estimation method based on engine torque estimation. Summary of the invention

[0006] The technical problem to be solved by the present invention is to overcome the problems in the existing vehicle mass estimation method that the torque and transmission ratio are difficult to obtain, which leads to the poor versatility of the method, and provide a vehicle mass estimation method based on engine torque estimation.

[0007] In order to solve the above technical problems, the present invention is implemented by adopting the following technical solution: comprising the following steps:

[0008] The first step is to determine the quality confidence

[0009] The vehicle quality confidence is defined as follows: The quality confidence is defined as a dimensionless number from 0 to 1, representing the degree of credibility of the quality estimate at the current moment, 1 represents credibility, 0 represents untrustworthiness, and the confidence is 0 when the vehicle stops for more than a certain limit. The quality confidence at the next moment is related to the quality confidence at the previous moment, the fuzzy logic membership, and the quality difference between the previous and next moments. If the quality difference between the current moment and the previous moment is large, it means that the quality estimate has not stabilized and is still in the calculation stage. At this time, the vehicle quality confidence decreases; if the quality difference between the current moment and the previous moment is small, it means that the estimated quality tends to stabilize and the confidence increases. At the same time, the invention of this article introduces the membership of fuzzy rules to correct the confidence. When the membership is small, it means that the current working conditions are not suitable for quality estimation. At this time, the membership has little effect on the confidence; when the membership is large, it means that the current working conditions are suitable for quality estimation. At this time, the membership has a greater impact on the confidence. When the vehicle quality confidence reaches a certain value, the current quality result can be considered reliable, and the current quality can be considered to remain unchanged until the next parking time exceeds the threshold value. The confidence calculation formula is as follows:

[0010]

[0011] Where, α(k) is the vehicle mass confidence at time k, unitless

[0012] β(k)——Parking time discrimination coefficient, when the parking time exceeds the limit, it takes 0, otherwise it takes 1, no unit

[0013] κ——empirical coefficient, unitless

[0014] F(k)——Fuzzy logic membership at time k, unitless

[0015] Δm(k)——The difference between the estimated mass at time k and the previous time, in kg

[0016] m max ——Mass difference threshold, in kg

[0017] Step 2: Data Acquisition

[0018] The vehicle's longitudinal speed, longitudinal acceleration, wheel speed, intake mass flow, air-fuel ratio and other information are obtained through the vehicle network bus and other sensors.

[0019] Step 3: Calculate engine torque and transmission ratio

[0020] Considering that some models cannot obtain engine torque data from the vehicle CAN bus, it is necessary to use the existing data to calculate the engine torque. Since the engine power can be expressed as:

[0021] P e =η e Q f H f (2)

[0022] Where P e ——Engine power, in W

[0023] η e ——Engine thermal efficiency, no unit

[0024] Q f ——Fuel mass flow rate, in kg / s

[0025] H f ——Heating value of fuel, in J / s

[0026] And the engine torque can be obtained by the engine power and speed:

[0027]

[0028] Where, T e ——Engine torque, in Nm

[0029] n e ——Engine speed, in r / min

[0030] Combining formula (2) and formula (3), we can get:

[0031]

[0032] In the formula, Q a ——Intake air mass flow, unit: kg / s

[0033] α AF ——Air-fuel ratio, no unit

[0034] Similarly, it is necessary to obtain the real-time transmission ratio data of the vehicle. Assuming that the tire slip of the vehicle is small and negligible, the vehicle speed calculation formula is:

[0035]

[0036] In the formula, v x ——Vehicle longitudinal speed, in m / s

[0037] n w ——Vehicle speed, unit: r / min

[0038] r——wheel radius, in m

[0039] The relationship between engine speed and wheel speed is:

[0040] n e =n w i g i0 (6)

[0041] In the formula, i g ——Transmission ratio, no unit

[0042] i0——Main reducer transmission ratio, no unit

[0043] Combining formula (5) and formula (6), the transmission ratio is:

[0044]

[0045] Step 4: Vehicle mass estimation working condition judgment based on fuzzy rules

[0046] Using dynamics to estimate vehicle mass requires considering the actual state of the vehicle. Fuzzy logic is needed to judge the current driving state of the vehicle. Under ideal working conditions, the formula is used to estimate the mass. The fuzzy rules are as follows:

[0047] First of all, considering that the main principle of the mass estimation algorithm is Newton's second law and the vehicle's longitudinal dynamic equilibrium equation, the ideal working condition requires a larger longitudinal force and longitudinal acceleration; similarly, in order to prevent the lateral movement from affecting the longitudinal movement, it is necessary to choose a smaller lateral acceleration range for mass estimation; the vehicle speed value is moderate, indicating that the car is in a continuous driving state and in the forward direction; the difference between the longitudinal force filter value and the original longitudinal force should not be too large; in order to avoid interference caused by sudden acceleration and deceleration, the longitudinal force change rate should not be too large.

[0048] Then the membership degree F(k) of the current working condition under the fuzzy rule is obtained.

[0049] Step 5: Quality estimation based on recursive least squares

[0050] According to the knowledge of vehicle kinematics, the data collected by the acceleration sensor includes both the vehicle longitudinal acceleration information and the slope information. The formula is as follows:

[0051]

[0052] In the formula, a x ——Longitudinal acceleration collected by the accelerometer, in m / s 2

[0053] δ——rotational mass conversion factor, unitless

[0054] g——acceleration due to gravity, in m / s 2

[0055] θ——slope, in rad

[0056] And the vehicle longitudinal dynamics equilibrium equation is:

[0057]

[0058] Where η t ——Transmission efficiency, no unit

[0059] C D ——Air resistance coefficient, unitless

[0060] A——windward area, in m 2

[0061] ρ——air density, kg / m 3

[0062] f——rolling resistance coefficient, no unit

[0063] Combined with formula (8), and considering that the slope angle is small in actual situations, it can be considered that mgfcosθ≈mgf, and the longitudinal dynamic equilibrium equation is converted into the following form:

[0064]

[0065] Convert formula (10) into the least squares form:

[0066]

[0067] In the formula, ——The estimated value of mass at time k

[0068] γ(k)——Gain coefficient at time k

[0069] P(k)——covariance matrix at time k

[0070] λ——Forgetting factor

[0071] Compared with the prior art, the present invention has the following beneficial effects:

[0072] 1. The vehicle mass estimation method based on engine torque estimation described in the present invention is to design a vehicle online mass estimation algorithm based on the basic data of the vehicle CAN bus, without using the engine torque signal and real-time transmission ratio information that cannot be obtained by some vehicle models. The method has the advantage of wide application;

[0073] 2. The vehicle mass estimation method based on engine torque estimation described in the present invention does not add other signals or devices on the basis of the traditional vehicle bus and sensor, which reduces the design burden and the cost of the vehicle online mass estimation algorithm;

[0074] 3. The vehicle mass estimation method based on engine torque estimation of the present invention can stabilize the mass estimation value after the confidence meets the requirements, thereby saving the computing power of the vehicle controller during the entire operation process;

[0075] 4. The vehicle mass estimation method based on engine torque estimation described in the present invention adopts the recursive least square method with a forgetting factor, which reduces the error caused by the drastic change of the vehicle force under complex working conditions and ensures the robustness and accuracy of the mass estimation value; BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The present invention will be further described below in conjunction with the accompanying drawings:

[0077] Figure 1 It is a logic diagram of the vehicle online quality estimation method of the present invention;

[0078] Figure 2 A membership function used in the fuzzy rule for working condition identification described in the present invention;

[0079] Figure 3 This is a diagram of the longitudinal dynamics analysis of the vehicle according to the present invention. DETAILED DESCRIPTION

[0080] The present invention is described in detail below in conjunction with the accompanying drawings:

[0081] The present invention discloses a vehicle mass estimation method based on engine torque estimation. The present invention utilizes basic data on the vehicle CAN bus and basic sensors such as accelerometers to achieve online estimation of the vehicle mass through the connection between the two. The method first obtains the longitudinal speed v of the vehicle x , longitudinal acceleration a x , wheel speed n w , intake air mass flow Q a , air-fuel ratio α AF Then, some non-basic bus data are calculated using this information, including the current engine torque T e And the gearbox ratio i g, and then use these data to judge the fuzzy rule membership of the current working condition, and further use the recursive least squares method with forgetting factor to estimate the quality. This quality estimation method only uses basic bus data and basic on-board sensors, so it can be widely used in different types of vehicles. At the same time, this method has sufficient robustness and accuracy, and can provide reasonable quality information for vehicle control, which can effectively improve the handling stability and fuel economy control of the whole vehicle. Next, the idea of ​​realizing vehicle quality estimation based on engine torque estimation is explained in detail.

[0082] See also Figure 1 The vehicle mass estimation method based on engine torque estimation of the present invention includes vehicle mass confidence calculation and judgment, engine torque and transmission ratio calculation, fuzzy rule membership solution and recursive least squares mass estimation. The vehicle mass estimation method based on engine torque estimation is described in detail step by step below.

[0083] The following steps are involved:

[0084] The first step is to determine the quality confidence

[0085] The vehicle quality confidence is a standard for measuring the credibility of the current quality information of the car. First, the vehicle quality confidence is defined as: the quality confidence is defined as a dimensionless number from 0 to 1, representing the credibility of the quality estimate at the current moment, 1 represents credibility, 0 represents untrustworthiness, and the confidence is 0 when the vehicle stops for more than a certain limit. The quality confidence at the next moment is related to the quality confidence at the previous moment, the fuzzy logic membership, and the quality difference between the previous and next moments. If the quality difference between the current moment and the previous moment is large, it means that the quality estimate has not stabilized and is still in the calculation stage. At this time, the vehicle quality confidence decreases; if the quality difference between the current moment and the previous moment is small, it means that the estimated quality tends to be stable and the confidence increases. At the same time, the present invention introduces the membership of fuzzy rules to correct the confidence. When the membership is small, it means that the current working condition is not suitable for quality estimation, and the membership has little effect on the confidence; when the membership is large, it means that the current working condition is suitable for quality estimation, and the membership has a greater impact on the confidence. When the vehicle quality confidence reaches a certain value, the current quality result can be considered reliable, and the current quality can be considered to remain unchanged until the next parking time exceeds the threshold value. The confidence calculation formula is as follows:

[0086]

[0087] Where, α(k) is the vehicle mass confidence at time k, unitless

[0088] β(k)——Parking time discrimination coefficient, when the parking time exceeds the limit, it takes 0, otherwise it takes 1, no unit

[0089] κ——empirical coefficient, unitless

[0090] F(k)——Fuzzy logic membership at time k, unitless

[0091] Δm(k)——The difference between the estimated mass at the current moment and the previous moment, in kg

[0092] m max ——Mass difference threshold, in kg

[0093] Step 2: Data Acquisition

[0094] Considering the restrictions of different automobile manufacturers on automobile bus data, it is necessary to use basic data to indirectly calculate and solve other data. In order to calculate the engine torque and current transmission ratio information, it is necessary to first obtain the vehicle longitudinal speed, longitudinal acceleration, wheel speed, intake mass flow, air-fuel ratio and other information through the vehicle network bus and other sensors.

[0095] Step 3: Calculate engine torque and transmission ratio

[0096] Considering that some models cannot obtain engine torque data from the vehicle CAN bus, it is necessary to use the existing data to calculate the engine torque. Since the engine power can be expressed as:

[0097] P e =η e Q f H f (13)

[0098] Where P e ——Engine power, in W

[0099] η e ——Engine thermal efficiency, no unit

[0100] Q f ——Fuel mass flow rate, in kg / s

[0101] H f ——Heating value of fuel, in J / s

[0102] And the engine torque can be obtained by the engine power and speed:

[0103]

[0104] Where, T e ——Engine torque, in Nm

[0105] n e ——Engine speed, in r / min

[0106] Combining formula (2) and formula (3), we can get:

[0107]

[0108] In the formula, Q a ——Intake air mass flow, unit: kg / s

[0109] α AF ——Air-fuel ratio, no unit

[0110] Similarly, it is necessary to obtain the real-time transmission ratio data of the vehicle. Assuming that the tire slip of the vehicle is small and negligible, the vehicle speed calculation formula is:

[0111]

[0112] In the formula, v x ——Vehicle longitudinal speed, in m / s

[0113] n w ——Vehicle speed, unit: r / min

[0114] r——wheel radius, in m

[0115] The relationship between engine speed and wheel speed is:

[0116] n e =n w i g i0 (17)

[0117] In the formula, i g ——Transmission ratio, no unit

[0118] i0——Main reducer transmission ratio, no unit

[0119] Combining formula (5) and formula (6), the transmission ratio is:

[0120]

[0121] Step 4: Vehicle mass estimation working condition judgment based on fuzzy rules

[0122] See also Figure 2 , the use of dynamics to estimate vehicle mass needs to consider the actual state of the vehicle, and the current driving state of the vehicle needs to be judged through fuzzy logic, and the mass estimation is performed using formulas under relatively ideal working conditions. The membership function in the fuzzy rule is determined by actual engineering experience, which has a great influence on the output results of the fuzzy rule. At the same time, the fuzzy rules are formulated as follows:

[0123] First of all, considering that the main principle of the mass estimation algorithm is Newton's second law and the vehicle's longitudinal dynamic equilibrium equation, the ideal working condition requires a larger longitudinal force and longitudinal acceleration; similarly, in order to prevent the lateral movement from affecting the longitudinal movement, it is necessary to choose a smaller lateral acceleration range for mass estimation; the vehicle speed value is moderate, indicating that the car is in a continuous driving state and in the forward direction; the difference between the longitudinal force filter value and the original longitudinal force should not be too large; in order to avoid interference caused by sudden acceleration and deceleration, the longitudinal force change rate should not be too large.

[0124] Then the membership degree F(k) of the current working condition under the fuzzy rule is obtained.

[0125] Step 5: Quality estimation based on recursive least squares

[0126] See also Figure 3 , establish the vehicle longitudinal dynamic balance equation. According to the knowledge of vehicle kinematics, the data collected by the acceleration sensor includes both the vehicle longitudinal acceleration information and the slope information. The formula is as follows:

[0127]

[0128] In the formula, a x ——Longitudinal acceleration collected by the accelerometer, in m / s 2

[0129] δ——rotational mass conversion factor, unitless

[0130] g——acceleration due to gravity, in m / s 2

[0131] θ——slope, in rad

[0132] And the vehicle longitudinal dynamics equilibrium equation is:

[0133]

[0134] Where η t ——Transmission efficiency, no unit

[0135] C D ——Air resistance coefficient, unitless

[0136] A——windward area, in m 2

[0137] ρ——air density, kg / m 3

[0138] f——rolling resistance coefficient, no unit

[0139] Combined with formula (8), and considering that the slope angle is small in actual situations, cosθ≈1, it can be considered that mgfcosθ≈mgf, and the longitudinal dynamic equilibrium equation is converted into the following form:

[0140]

[0141] Convert formula (10) into the least squares form:

[0142]

[0143] In the formula, ——The estimated value of mass at time k

[0144] γ(k)——Gain coefficient at time k

[0145] P(k)——covariance matrix at time k

[0146] λ——Forgetting factor

[0147] Then the estimated value of the car mass at the current moment is solved.

Claims

1. A method for estimating vehicle mass based on engine torque estimation, characterized in that The following steps are involved: The first step is to determine the quality confidence The vehicle quality confidence is defined as follows: The quality confidence is defined as a dimensionless number from 0 to 1, representing the credibility of the quality estimate at the current moment, 1 represents credibility, 0 represents untrustworthiness, and the confidence is 0 when the vehicle stops for more than a certain limit; the quality confidence at the next moment is related to the quality confidence at the previous moment, the fuzzy logic membership, and the quality difference between the previous and next moments; if the quality difference between the current moment and the previous moment is large, it means that the quality estimate has not stabilized and is still in the calculation stage. At this time, the vehicle quality confidence decreases. If the quality difference between the current moment and the previous moment is When it is smaller, it means that the estimated quality tends to be stable and the confidence increases; at the same time, the invention of this paper introduces the membership of fuzzy rules to correct the confidence. When the membership is small, it means that the current working condition is not suitable for quality estimation. At this time, the membership has a small impact on the confidence. When the membership is large, it means that the current working condition is suitable for quality estimation. At this time, the membership has a greater impact on the confidence. When the vehicle quality confidence reaches a certain value, it can be considered that the current quality result is reliable, and the current quality can be considered to remain unchanged until the next parking time exceeds the threshold value. The confidence calculation formula is as follows: Where, α(k) is the vehicle mass confidence at time k, unitless β(k)——Parking time discrimination coefficient, when the parking time exceeds the limit, it takes 0, otherwise it takes 1, no unit κ——empirical coefficient, unitless F(k)——Fuzzy logic membership at time k, unitless Δm(k)——The difference between the estimated mass at the current moment and the previous moment, in kg m max ——Mass difference threshold, in kg Step 2: Data Acquisition Obtain vehicle longitudinal speed, longitudinal acceleration, wheel speed, intake mass flow, air-fuel ratio and other information through the vehicle network bus and other sensors; Step 3: Calculate engine torque and transmission ratio Considering that some models cannot obtain engine torque data from the vehicle CAN bus, it is necessary to use the existing data to calculate the engine torque; since the engine power can be expressed as: P e =η e Q f H f (2) Where P e ——Engine power, in W η e ——Engine thermal efficiency, no unit Q f ——Fuel mass flow rate, in kg / s H f ——Heating value of fuel, in J / s And the engine torque can be obtained by the engine power and speed: Where, T e ——Engine torque, in Nm n e ——Engine speed, in r / min Combining formula (2) and formula (3), we can get: In the formula, Q a ——Intake air mass flow, unit: kg / s α AF ——Air-fuel ratio, no unit Similarly, it is necessary to obtain the real-time transmission ratio data of the vehicle. Assuming that the tire slip of the vehicle is small and negligible, the vehicle speed calculation formula is: In the formula, v x ——Vehicle longitudinal speed, in m / s n w ——Vehicle speed, unit: r / min r——wheel radius, in m The relationship between engine speed and wheel speed is: n e =n w I g i0 (6) In the formula, i g ——Transmission ratio, no unit i0——Main reducer transmission ratio, no unit Combining formula (5) and formula (6), the transmission ratio is: Step 4: Vehicle mass estimation working condition judgment based on fuzzy rules Using dynamics to estimate vehicle mass requires considering the actual state of the vehicle. Fuzzy logic is needed to judge the current driving state of the vehicle. Under ideal working conditions, the formula is used to estimate the mass. The fuzzy rules are as follows: First, considering that the main principle of the mass estimation algorithm is Newton's second law and the vehicle longitudinal dynamics equilibrium equation, the ideal working condition requires a larger longitudinal force and longitudinal acceleration. Similarly, in order to prevent the lateral movement from affecting the longitudinal movement, it is necessary to select a smaller lateral acceleration range for mass estimation. The vehicle speed value is moderate, indicating that the car is in a continuous driving state and in the forward direction. The difference between the longitudinal force filter value and the original longitudinal force should not be too large. In order to avoid interference caused by rapid acceleration and deceleration, the longitudinal force change rate should not be too large. Then the membership degree F(k) of the current working condition under the fuzzy rule is obtained; Step 5: Quality estimation based on recursive least squares According to the knowledge of vehicle kinematics, the data collected by the acceleration sensor includes both the vehicle longitudinal acceleration information and the slope information. The formula is as follows: In the formula, a x ——Longitudinal acceleration collected by the accelerometer, in m / s 2 δ——rotational mass conversion factor, unitless g——acceleration due to gravity, in m / s 2 θ——slope, in rad And the vehicle longitudinal dynamics equilibrium equation is: Where η t ——Transmission efficiency, no unit C D ——Air resistance coefficient, unitless A——windward area, in m 2 ρ——air density, kg / m 3 f——rolling resistance coefficient, no unit Combined with formula (8), and considering that the slope angle is small in actual situations, it can be considered that mgfcosθ≈mgf, and the longitudinal dynamic equilibrium equation is converted into the following form: Convert formula (10) into the least squares form: In the formula, ——The estimated value of mass at time k γ(k)——Gain coefficient at time k P(k)——covariance matrix at time k λ——Forgetting factor Then the estimated value of the car mass at the current moment is solved.

Images