Commercial vehicle adaptive roll-over warning method based on fuzzy rules

Abstract

The application provides a commercial vehicle self-adaptive rollover early warning method based on fuzzy rules, which combines the absolute values of the steering wheel angular velocity and the vehicle longitudinal speed, which are more front signals, with the absolute value of the load lateral transfer rate, establishes a fuzzy rule based on the type I fuzzy logic with the load lateral transfer rate, the vehicle speed signal and the steering wheel angular velocity, performs self-adaptive rollover early warning on the commercial vehicle by means of fuzzification, fuzzy reasoning and de-fuzzification, and outputs a risk coefficient; the steering wheel angular velocity can reflect the steering intention of the driver, so that the fast early warning and the steering working condition self-adaptation are realized; the actual vehicle speed is represented as the vehicle load related by means of the equivalent vehicle speed, the domain scaling of the vehicle speed membership function is realized, and the rule is adapted to the vehicle load change. The application can realize fast, accurate and stable commercial vehicle rollover early warning under different steering working conditions and different load working conditions.

Description

Technical Field

[0001] This invention relates to the field of vehicle driving control safety systems, and in particular to an adaptive rollover warning method for commercial vehicles based on fuzzy rules. Background Technology

[0002] As a major component of the modern road transport system, commercial vehicles play an irreplaceable role in logistics. Compared to passenger cars, commercial vehicles have a higher center of gravity, longer wheelbase and track width, and greater sprung mass, making them far more prone to rollover accidents. If known vehicle signals can be used to estimate the current vehicle state and assess the current rollover risk, the driving safety of commercial vehicles can be effectively improved.

[0003] The current mainstream method for rollover warning of commercial vehicles is based on predicting the vehicle state using the time-to-rollover (TTR) and determining the risk level according to a risk threshold. Patent CN121106326A discloses a semi-trailer rollover warning method and system based on dynamic thresholds. This method comprehensively considers the vehicle's roll angle and load lateral transfer rate (LTR), and uses a TTR algorithm to linearly predict the state of the vehicle at the next moment based on the vehicle's state at the previous moment. The drawback of this method is that linear extrapolation can introduce significant errors for nonlinear changes in the predicted signal. Typically, vehicle state variables such as LTR and acceleration are difficult to change linearly, leading to unknown advances or lags in the prediction time, ultimately causing false alarms or missed alarms. CN120003461A discloses an active anti-rollover control system and method for electric heavy-duty trucks based on joint warning. It calculates the rollover time (TTR) value using the load lateral transfer rate calculated from the vehicle dynamics model and the zero-moment point, and selects an appropriate lower-level control module based on the comparison between the TTR and a set threshold. Under aperiodic operating conditions, the linear prediction of the TTR algorithm cannot guarantee the accuracy of the upper-level control target generation, and the motion behavior of the lower-level actuators may lead to unpredictable vehicle motion. That is, even if there was no risk of rollover or instability initially, the inaccuracy of the upper-level control target can cause the vehicle's stability to decrease after the lower-level mechanism executes.

[0004] Another type of rollover warning method directly assesses risk based on the current vehicle status signal combined with a predetermined threshold. Patent CN119796223A discloses a rollover risk warning method, device, medium, and equipment for liquid tanker vehicles. It uses the vehicle's total weight, mass, roll angle, and turning radius as inputs, performs fuzzification, and defuzzifies according to fuzzy rules to output a rollover risk value. The drawback of this method is that it can only determine the current vehicle status; the output rollover risk value lacks advance warning, thus losing its warning significance. Patent CN117261874A discloses a vehicle rollover risk warning method based on adaptive unscented Kalman filtering. It uses an unscented Kalman filter to predict the distance from the center of mass to the roll axis, calculates the LTR, and thus determines the risk level. This method can improve the accuracy of LTR calculation, but it also has a lag. This method is more suitable for vehicle status monitoring than for early warning.

[0005] In summary, the current vehicle rollover warning methods suffer from insufficient speed, accuracy, and stability. Regarding speed, the warning system needs to provide a certain lead time to allow sufficient response time for lower-level control modules. In terms of accuracy and stability, the system needs to be able to adapt to arbitrary linear and nonlinear changes in the input signal. Therefore, there is an urgent need for a rollover warning method that can adapt to complex operating conditions, possesses high precision, and exhibits strong robustness. Summary of the Invention

[0006] To address the above problems, this invention provides an adaptive rollover warning method for commercial vehicles under steering conditions based on fuzzy rules, as follows:

[0007] Step 1: Collect vehicle speed signal, steering wheel angular velocity and four-wheel vertical load, and calculate the load lateral transfer rate based on the four-wheel vertical load;

[0008] Step 2: Calculate the membership degree of each fuzzy subset based on the membership functions of vehicle speed, steering wheel angular velocity, and load lateral transfer rate, and infer the fuzzy subset to which it belongs.

[0009] In the vehicle speed membership function, fuzzy subsets corresponding to zero, small-medium, medium, medium-large, and large are set according to the range of variation of the vehicle speed signal; in the load lateral transfer rate membership function, fuzzy subsets corresponding to very small, small, medium, large, and very large are set within the universe of discourse of the absolute value of the load lateral transfer rate |LTR|; in the steering wheel angular velocity membership function, fuzzy subsets corresponding to very small, small, medium, large, and very large are set within the universe of discourse of the absolute value of the steering wheel angular velocity |dθ|. sw / dt| Defines fuzzy subsets corresponding to small, medium, and large within the domain of discourse;

[0010] Step 3: Based on the fuzzy subsets assigned by vehicle speed, steering wheel angular velocity, and load lateral transfer rate, the membership degrees of the zero, low, medium, and high fuzzy subsets corresponding to the risk coefficients are obtained by fuzzy rule reasoning; the fuzzy rules are specified in the table below;

[0011]

[0012] Step 4: Obtain the accurate risk coefficient through defuzzification, and use it to provide early warning of vehicle rollover.

[0013] Preferably, in step one, the equivalent vehicle speed is used as the input for the vehicle speed, and the expression for the equivalent vehicle speed is as follows:

[0014] (1)

[0015] In equation (1), V xm For the equivalent vehicle speed, m h For cargo weight, v x This refers to the actual vehicle speed.

[0016] Specifically, the formula for calculating the lateral load transfer rate based on the vertical load of the four wheels in step one is as follows:

[0017] (2)

[0018] In equation (2): F z11 F is the vertical force on the left front wheel. z21 F is the vertical force on the left rear wheel. z12 F is the vertical force on the right front wheel. z22 This is the vertical force on the right rear wheel.

[0019] Preferably, the membership function in step two is represented by a trapezoidal function; the standard form of the trapezoidal function used is as follows:

[0020] (3)

[0021] In equation (3): m is the membership degree, a is the x-coordinate of the first endpoint, b is the x-coordinate of the second endpoint, c is the x-coordinate of the third endpoint, and d is the x-coordinate of the fourth endpoint;

[0022] If equation (3) is rewritten when the x-coordinates of the first and second endpoints coincide or the x-coordinates of the third and fourth endpoints coincide, the following is true:

[0023] (4)

[0024] In equation (4), m1 is the membership degree when the horizontal coordinates of the first and second endpoints coincide, and m2 is the membership degree when the horizontal coordinates of the third and fourth endpoints coincide.

[0025] Further optimized, the endpoint coordinates of the vehicle speed membership function in step two.

[0026] (5)

[0027] Load lateral transfer rate membership function endpoint coordinates

[0028] (6)

[0029] Steering wheel angular velocity membership function endpoint coordinates

[0030] (7)

[0031] Risk coefficient membership endpoint coordinates

[0032] (8)

[0033] Preferably, in step four, the precise risk coefficient in the interval [0,1] is divided into five risk levels with an interval length of 0.2: "Safe" [0,0.2), "Caution" [0.2,0.4), "Caution" [0.4,0.6), "Warning" [0.6,0.8), and "Danger" [0.8,1], thereby providing a warning for vehicle rollover.

[0034] The beneficial effects of this invention are:

[0035] This invention combines two more advanced signals—the absolute value of steering wheel angular velocity and vehicle longitudinal velocity—with the absolute value of load lateral transfer rate, and designs an adaptive rollover warning module based on Type I fuzzy logic to output a risk coefficient. In the fuzzy rules, the load lateral transfer rate and vehicle speed signal directly affect the warning intensity, while steering wheel angular velocity affects the warning timing. By introducing steering wheel angular velocity, the driver's steering intention can be reflected, thereby achieving rapid warning and adaptive steering conditions. To avoid introducing a new membership function that would make the fuzzy rules too large, the actual vehicle speed is represented as being related to the vehicle load by using an equivalent vehicle speed, thus scaling the universe of discourse of the vehicle speed membership function and making the rules adaptable to changes in vehicle load.

[0036] The method of this invention can achieve fast, accurate and stable rollover warning for commercial vehicles under different steering conditions and different load conditions. Attached Figure Description

[0037] Figure 1 It's based on the principle of rollover warning;

[0038] Figure 2 It is the vehicle's trajectory;

[0039] Figure 3 These are membership functions: (a) vehicle speed membership function; (b) load lateral transfer rate membership function; (c) steering wheel angular velocity membership function; (d) risk coefficient membership function;

[0040] Figure 4These are fuzzy inference simulation results; in the figure: V represents vehicle speed, LTR represents load lateral transfer rate, Steering_rate represents steering wheel angular velocity, and risk is the risk coefficient;

[0041] Figure 5 It is based on the principle of rollover time warning;

[0042] Figure 6 The TTR warning effects are: (a) TTR linear input warning effect; (b) TTR non-linear input warning effect.

[0043] Figure 7 The steering wheel angle and angular velocity are: (a) steering wheel angle and angular velocity under the fishhook condition; (b) steering wheel angle and angular velocity under the sine wave condition.

[0044] Figure 8 The comparison of rollover warning effects is as follows: (a) TTR warning effect under hook working conditions; (b) TTR warning effect under sinusoidal working conditions; (c) rollover warning effect under hook working conditions; (d) rollover warning effect under sinusoidal working conditions.

[0045] Figure 9 The warning effects are: (a) warning effect for fully loaded fishhook working conditions; (b) warning effect for fully loaded sinusoidal working conditions. Detailed Implementation

[0046] The implementation steps of the rollover warning method of the present invention are as follows:

[0047] according to Figure 1 As can be seen, the core idea of ​​this invention is to comprehensively evaluate the vehicle rollover state using multiple state variables based on fuzzy rules. The external inputs to the system are vehicle speed, vehicle cargo mass, vertical load on the four wheels, and steering wheel angular velocity. These state variables can be estimated online based on the vehicle dynamics model or measured by sensors, and are considered known variables in this invention.

[0048] Load lateral transfer rate (LTR) is commonly used as an evaluation index for assessing vehicle rollover risk. The closer |LTR| is to 1, the greater the likelihood of a rollover. While LTR can roughly characterize the critical state of rollover risk when a vehicle is about to roll over, it has limitations due to its lag and suitability for assessment under complex conditions. Steering wheel angular velocity typically reflects the driver's steering intention; a larger angular velocity indicates an emergency turn, while a smaller angular velocity indicates a gradual turn. This invention, guided by driver operating characteristics, combines the absolute value of steering wheel angular velocity and vehicle longitudinal velocity—two more advanced signals—with |LTR|. Based on Type I fuzzy logic, an adaptive rollover warning module is designed to output a risk coefficient. The basic principle of the rollover warning module is as follows: Figure 1 As shown.

[0049] The early warning module comprises a three-layer architecture: an input fuzzification module, fuzzy rules, and an output defuzzification module.

[0050] Step 001: The load lateral transfer rate needs to be calculated using the four-wheel vertical load according to the given LTR calculation formula, as follows:

[0051] (1)

[0052] In equation (1): F z11 F is the vertical force on the left front wheel. z21 F is the vertical force on the left rear wheel. z12 F is the vertical force on the right front wheel. z22 This is the vertical force on the right rear wheel.

[0053] Step 002: When using actual vehicle speed, LTR, and steering wheel angular velocity as inputs to the fuzzification module, the impact of vehicle load cannot be reflected in the rollover warning module. If a new load membership function is added, assuming that the membership function includes 3 fuzzy subsets, the number of fuzzy rules will increase from the original 75 to 225. At this point, the difficulty of designing fuzzy rules increases dramatically, and the computational efficiency is greatly reduced.

[0054] The direct factor affecting the lateral transfer of vehicle load is the lateral acceleration generated by the vehicle during steering. Therefore, we analyze the motion state of the vehicle during steering, and the motion trajectory of the vehicle during steering is given below. Figure 2 As shown.

[0055] exist Figure 2 In the equation, the change in the vehicle's velocity component along the y-axis can be expressed as follows:

[0056] (2)

[0057] In equation (2): v y Let v be the lateral velocity of the vehicle. y Let v be the change in lateral velocity at time ∆t. x For vehicle speed, ∆v x ∆t represents the change in vehicle speed at time ∆t, and ∆θ represents the change in steering angle.

[0058] Considering that ∆θ is very small, then sin∆θ≈∆θ, cos∆θ≈1. Ignoring the second-order differential, the above equation can be transformed into:

[0059] (3)

[0060] Vehicle lateral acceleration Expressed as the change in velocity divided by the unit time ∆t:

[0061] (4)

[0062] Centrifugal force when a vehicle turns Represented as:

[0063] (5)

[0064] In equations (3) to (5): m s For the sprung mass of the vehicle, This refers to the vehicle's steering angular velocity.

[0065] Analysis of the above formula shows that the centrifugal force during vehicle turning is positively correlated with both the sprung mass and the longitudinal vehicle speed. When the centrifugal force is constant, this invention assumes a variable that can describe the longitudinal vehicle speed under different load conditions. According to Article 78 of the "Regulations for the Implementation of the Road Traffic Safety Law of the People's Republic of China," the maximum speed of freight vehicles traveling on highways shall not exceed 100 km / h, and the maximum speed shall not exceed 80 km / h when overloaded by 30%. This invention correlates the vehicle speed signal with the load of commercial vehicles, proposing the concept of equivalent vehicle speed. Using the equivalent vehicle speed as the input to the module described in this invention can be equivalent to scaling the domain of the vehicle speed membership function. The expression for the equivalent vehicle speed is as follows:

[0066] (6)

[0067] In equation (6): V xm For the equivalent vehicle speed, m h For cargo weight, v x For vehicle speed

[0068] The equivalent vehicle speed calculation module described above is designed based on the vehicle's rated load capacity in Trucksim. This commercial vehicle has an unloaded weight of 4455 kg and a fully loaded weight of 4 tons. Therefore, the equivalent vehicle speed is specified to increase with increasing load; a fully loaded speed of 80 km / h is equivalent to an unloaded speed of 100 km / h. This speed equivalence method is equivalent to reducing the domain of the vehicle speed membership by a factor of (1, 1.25) based on the load, thus reflecting the vehicle's load status in the speed signal.

[0069] Step 003: Unlike classic Boolean logic, fuzzy logic quantifies a precise input value into a fuzzy concept. To achieve numerical fuzzification, a corresponding membership function needs to be selected and designed. Considering the system's real-time requirements, this invention uses a trapezoidal function as the membership function, such as... Figure 3 As shown.

[0070] According to the speed limit for commercial vehicles stipulated in my country's Road Traffic Safety Law, the domain of the vehicle speed membership function is limited to [0, 100] km / h. When using vehicle speed signals acquired by sensors, it is necessary to classify the speed levels. Typically, the concepts of "low speed," "medium speed," and "high speed" are only fuzzy concepts. This invention sets five fuzzy subsets in the vehicle speed membership function: ZO, MS, M, MB, and B, corresponding to the Chinese meanings of "zero," "small-medium," "medium," "medium-large," and "large." The non-zero regions of the ZO and B fuzzy subsets are of equal length, as are the non-zero regions of the MS, M, and MB fuzzy subsets. In this embodiment, the fuzzy subsets ZO, MS, M, MB, and B are set as follows:

[0071] When the vehicle speed is set to [0, 12.5] km / h in the ZO fuzzy subset, the membership function is always 1; within the range of (12.5, 22.5) km / h, the membership function linearly decreases to 0, and the membership function is always 0 when the vehicle speed is greater than or equal to 22.5 km / h.

[0072] In the MS fuzzy subset, when the vehicle speed is less than or equal to 2.5 km / h, the membership degree is always 0; in the range of (2.5, 22.5) km / h, the membership degree increases linearly from 0 to 1; in the range of [22.5, 27.5] km / h, the membership degree is always 1; when the vehicle speed is (27.5, 47.5) km / h, the membership degree decreases linearly from 1 to 0; and when the vehicle speed is greater than or equal to 47.5 km / h, the membership degree is always 0.

[0073] In the M fuzzy subset, the membership degree is always 0 when the vehicle speed is less than or equal to 27.5 km / h; in the range of (27.5, 47.5) km / h, the membership degree increases linearly from 0 to 1; in the range of [47.5, 52.5] km / h, the membership degree is always 1; when the vehicle speed is (52.5, 72.5) km / h, the membership degree decreases linearly from 1 to 0; and when the vehicle speed is greater than or equal to 72.5 km / h, the membership degree is always 0.

[0074] In the MB fuzzy subset, the membership degree is always 0 when the vehicle speed is less than or equal to 52.5 km / h; in the range of (52.5, 72.5) km / h, the membership degree increases linearly from 0 to 1; in the range of [72.5, 77.5] km / h, the membership degree is always 1; when the vehicle speed is (77.5, 97.5) km / h, the membership degree decreases linearly from 1 to 0; and when the vehicle speed is greater than or equal to 97.5 km / h, the membership degree is always 0.

[0075] In the B fuzzy subset, when the vehicle speed is less than or equal to 77.5 km / h, the membership degree is always 0; in the range of (77.5, 87.5) km / h, the membership degree increases linearly from 0 to 1; and in the range of [87.5, 100] km / h, the membership degree is always 1.

[0076] Based on the settings of the two fuzzy subsets mentioned above, if the vehicle speed is 10 km / h, the membership degree of the corresponding fuzzy subset ZO is 1, and the membership degree of the fuzzy subset MS is 0.3. In subsequent fuzzy rules, the membership degrees of the two fuzzy subsets will be activated and used as the pre-input of the fuzzy rules. At this time, the fuzzy rules will output the judgment result based on the two membership degrees. This invention sets the union of the two judgment results, that is, the maximum value, to ensure the warning strength. The setting of the remaining fuzzy subsets in the vehicle speed membership function is similar and will not be repeated here.

[0077] Step 004: The actual LTR range is [-1, 1]. To reduce the computational cost of fuzzy rules, this invention uses the absolute value of the LTR. As input, the universe of discourse for the load lateral transfer rate membership function is defined as [0,1]. Five fuzzy subsets are defined as: SS, S, M, B, and BB, which correspond to "very small", "small", "medium", "large", and "very large" in Chinese, respectively. The non-zero regions of the SS and BB fuzzy subsets are of equal length, and the non-zero regions of the S and B fuzzy subsets are of equal length. In this embodiment, the fuzzy subsets SS, S, M, B, and BB are defined as follows:

[0078] Setting in SS fuzzy subset When the membership degree is in the range [0, 0.2], it is always 1; in the range (0.2, 0.3), the membership degree decreases linearly to 0; when When the membership degree is ≥0.3, the membership degree is always 0.

[0079] Set in the S fuzzy subset When the membership value is ≤0.2, the membership degree is always 0; in the range (0.2, 0.3), the membership degree increases linearly from 0 to 1, and in the range [0.3, 0.4], it is always 1; when When the membership is (0.4, 0.5), the membership degree decreases linearly to 0; when When the membership degree is ≥0.5, the membership degree is always 0.

[0080] Set in the M fuzzy subset When the membership value is ≤0.4, the membership degree is always 0; within the range (0.4, 0.45), the membership degree increases linearly from 0 to 1, and remains constant at 1 within the range [0.45, 0.55]. When the membership is (0.55, 0.6), the membership decreases linearly to 0; when When the membership degree is ≥0.6, the membership degree is always 0.

[0081] Set in the B fuzzy subset When the membership value is ≤0.5, the membership degree is always 0; within the range (0.5, 0.6), the membership degree increases linearly from 0 to 1, and remains constant at 1 within the range [0.6, 0.7]. When the membership is (0.7, 0.8), the membership degree decreases linearly to 0; when When the membership degree is ≥0.8, the membership degree is always 0.

[0082] Setting in the BB fuzzy subset When the membership value is ≤0.7, the membership degree is always 0; in the range of (0.7,0.8), the membership degree increases linearly from 0 to 1, and in the range of [0.8,1], it is always 1.

[0083] Step 005: Design the steering wheel angular velocity membership function, setting the function input to the absolute value of the steering wheel angular velocity. Regarding the selection of the universe of discourse, this invention sets the universe of discourse to [0, 800] deg / s based on the steering wheel angular velocity range under Trucksim's fishing hook condition. Three fuzzy subsets are defined in the steering wheel angular velocity membership function: S, M, and B, corresponding to the Chinese meanings "small," "medium," and "large," respectively. The non-zero region lengths of each fuzzy subset are equal. In this embodiment, the fuzzy subsets S, M, and B are defined as follows:

[0084] When the steering wheel angular velocity is set to the range of [0, 133.3] deg / s in the S fuzzy subset, the membership degree is always 1; within the range of (133.3, 400), the membership degree decreases linearly from 1 to 0; when the steering wheel angular velocity is greater than or equal to 400 deg / s, the membership degree is always 0.

[0085] In the M fuzzy subset, the membership degree is always 0 when the steering wheel angular velocity is less than or equal to 132.9 deg / s; it increases linearly from 0 to 1 when the steering wheel angular velocity is in the range of (132.9, 266.2) deg / s; it remains 1 when the steering wheel angular velocity is in the range of [266.2, 532.9] deg / s; it decreases linearly from 1 to 0 when the steering wheel angular velocity is in the range of (532.9, 666.2) deg / s; and it remains 0 when the steering wheel angular velocity is greater than or equal to 666.2 deg / s.

[0086] In the M fuzzy subset, when the steering wheel angular velocity is less than or equal to 400 deg / s, the membership degree is always 0; when it is (400, 666.7) deg / s, the membership degree increases linearly from 0 to 1; when the steering wheel angular velocity is [666.7, 800] deg / s, the membership degree is always 1.

[0087] Step 006: The risk coefficient membership function directly determines the final output risk coefficient. To facilitate comparison with the LTR, this invention sets the universe of discourse of the risk coefficient membership function to [0,1], referencing the universe of discourse of the LTR. Four fuzzy subsets are defined in the risk coefficient membership function: ZO, S, M, and B, corresponding to the Chinese meanings of "zero," "low," "medium," and "high." The non-zero region lengths of the fuzzy subsets ZO and B are equal, and the non-zero region lengths of the fuzzy subsets S and M are equal. In this embodiment, the fuzzy subsets ZO, S, M, and B are defined as follows:

[0088] In the ZO fuzzy subset, if the risk coefficient is in the range of [0, 0.2], the membership degree is always 1; when it is in the range of (0.2, 0.3), the membership degree decreases linearly from 1 to 0; when the risk coefficient is greater than or equal to 0.3, the membership degree is always 0.

[0089] In the S fuzzy subset, if the risk coefficient is less than or equal to 0.2, the membership degree is always 0; when the risk coefficient is in the range of (0.2, 0.3), the membership degree increases linearly from 0 to 1; when it is in the range of [0.3, 0.4], the membership degree is always 1; when it is in the range of (0.4, 0.5), the membership degree decreases linearly from 1 to 0; if the risk coefficient is greater than or equal to 0.5, the membership degree is always 0.

[0090] In the M fuzzy subset, if the risk coefficient is less than or equal to 0.4, the membership degree is always 0; when the risk coefficient is in the range of (0.4, 0.6), the membership degree increases linearly from 0 to 1; when it is in the range of [0.6, 0.7], the membership degree is always 1; when it is in the range of (0.7, 0.8), the membership degree decreases linearly from 1 to 0; if the risk coefficient is greater than or equal to 0.8, the membership degree is always 0.

[0091] In the fuzzy subset B, if the risk coefficient is less than or equal to 0.7, the membership degree is always 0; when the risk coefficient is in the range of (0.7, 0.8), the membership degree increases linearly from 0 to 1; when it is in the range of [0.8, 1], the membership degree is always 1.

[0092] Step 007: In the fuzzy toolbox of Matlab, establish the above membership functions and implement the 75 fuzzy rules in Table 1. Set the weight of each fuzzy rule to 1, and connect the preconditions with AND. That is, the risk coefficient membership function must simultaneously meet the fuzzy subsets contained in the above three different membership functions to determine which fuzzy subset it is contained in.

[0093] Table 1

[0094]

[0095] Table 1 consists of three side-by-side sub-tables, each corresponding to the steering wheel angular velocity (|dθ). swThe membership degree of / dt| is {small (S), medium (M), large (B)}. The first column of the table is the membership degree of the absolute value of the load lateral transfer rate (|LTR|), which is {very small (SS), small (S), medium (M), large (B), very large (BB)}. The second row is the vehicle speed (V). xm The membership degrees of the risk coefficients are {Z0 (ZO), Small to Medium (MS), Medium (M), Medium to Large (MB), Large (B)}. The 5×5 table below shows the membership degrees of the risk coefficients, which are {Z0 (ZO), Low (S), Medium (M), High (B)}.

[0096] Since the load lateral transfer rate is a crucial indicator directly reflecting vehicle lateral stability, it carries the highest weight in this fuzzy rule. Vehicle speed amplifies rollover risk; with the same steering wheel input, higher vehicle speed results in greater load lateral transfer during steering. Steering wheel angular velocity directly reflects the driver's steering intention, and sudden steering maneuvers are a common cause of rollovers. This paper will now explain the fuzzy rule in detail, using the membership degree of the load lateral transfer rate as the primary variable and steering wheel angular velocity and vehicle speed as secondary variables.

[0097] When the membership degree of |LTR| is "very small", it indicates that there is basically no load transfer in the current vehicle; at the same time, if |dθ sw If the membership degree of / dt| is "small", the risk coefficient membership degree remains "zero" as the vehicle speed increases, meaning the vehicle is in a safe state and no warning is needed; if |dθ| sw / dt|Membership degree is "medium". As vehicle speed increases, the risk coefficient membership degree gradually changes from "zero" to "small". At this point, the vehicle is considered to have low risk at high speed; if |dθ sw / dt| With a membership degree of "large", as the vehicle speed increases, the membership degree of the risk coefficient transitions from "zero" to "medium", at which point the vehicle is considered to have a medium risk at high speed.

[0098] When the membership degree of |LTR| is "small", it indicates that there is a small amount of load transfer in the current vehicle; at the same time, if |dθ sw / dt| has a membership degree of "small". As vehicle speed increases, the membership degree of the risk coefficient changes from "zero" to "small". At this point, the vehicle only has low risk at "medium-high" and "high" speeds; if |dθ sw The membership degree of / dt| is "medium". As the vehicle speed increases, the membership degree of the risk coefficient changes from "zero" to "medium", meaning that the vehicle is considered to have low risk at "medium" and "medium-high" speeds, and medium risk at "high" speeds; if |dθ sw / dt| has a membership degree of "large". As the vehicle speed increases, the membership degree of the risk coefficient changes from "zero" to "large", which means that there is a high risk when the vehicle makes a rapid turn at "medium-large" speed.

[0099] When the membership degree of |LTR| is "medium", it indicates that the current vehicle exhibits a moderate degree of load transfer; simultaneously, if |dθ sw / dt| has a membership degree of "small". As vehicle speed increases, the membership degree of the risk coefficient changes from "zero" to "medium", meaning that the vehicle is considered to have a medium risk at speeds of "medium, medium-high, high". If |dθ sw / dt| has a membership degree of "medium". As vehicle speed increases, the membership degree of the risk coefficient changes from "zero" to "large", meaning that the vehicle is considered to have a high risk at "high" speeds; if |dθ sw / dt| has a membership degree of "large". As the vehicle speed increases, the membership degree of the risk coefficient changes from "small" to "large". At this time, the vehicle is always under risk warning, and even at medium speeds, there is a high risk when making rapid turns.

[0100] When the membership degree of |LTR| is "large", it indicates that there is a significant load transfer in the current vehicle; at this time, any improper steering operation will cause a serious rollover accident, therefore only when |dθ| is the membership degree of LTR is "large". sw / dt| When the membership degree is "small, medium" and the steering speed is extremely low, the risk level is set to medium; otherwise, it is set to high risk.

[0101] When the membership degree of |LTR| is "very large", it indicates that the vehicle is about to overturn; during this stage, the warning module will constantly output a high risk.

[0102] Step 008: Import the designed fuzzy rules into the Matlab workspace. In Simulink, use the Fuzzylogic module to call the fuzzy rules established in Step 007. Configure the interface between Trucksim and Simulink, with Trucksim outputting the vertical load of the four wheels, longitudinal vehicle speed, and steering wheel angular velocity. After taking the absolute value or scaling these three signals, input them to the Fuzzylogic module via the Mux bus. The module performs fuzzification, fuzzy inference, and defuzzification sequentially. Set the corresponding steering conditions in Trucksim and run the program in Simulink to implement a rollover warning.

[0103] In this embodiment, the vehicle speed is set to 40 km / h, the load lateral transfer rate is 0.2, and the steering wheel angular velocity is 250 deg / s. The fuzzification process of the fuzzy logic module can be viewed as drawing a perpendicular line at a certain value in the universe of discourse, and the ordinate of the intersection point with each fuzzy subset in the membership function represents the membership degree of the current state. According to the vehicle speed membership function, the vehicle speed has a membership degree of 0.375 in the MS fuzzy subset and 0.625 in the M fuzzy subset. According to the load lateral transfer rate membership function, the LTR has a membership degree of 1 in the SS fuzzy subset. According to the steering wheel angular velocity membership function, the steering wheel angular velocity has a membership degree of 0.5 in the S fuzzy subset and 1 in the M fuzzy subset. Arrange the above membership degrees in the order of vehicle speed, LTR, and steering wheel angular velocity, including the following four cases: "MS-0.375, SS-1, S-0.5"; "MS-0.375, SS-1, M-1"; "M-0.625, SS-1, S-0.5"; "M-0.625, SS-1, M-1".

[0104] Inference is performed based on the fuzzy rules established in step 007 (see Table 1). When the membership degrees of the above three variables all meet the conditions, the fuzzy subsets output by the activated fuzzy rules are "ZO, ZO, ZO, S". The activation intensity is obtained by using the Min function to take the minimum value of each membership function, which is "0.375, 0.375, 0.5, 0.625". Then, the three ZO fuzzy subsets are aggregated by using the Max function to take the maximum value. The aggregated fuzzy subsets are "ZO, S", with activation intensities of "0.5, 0.625". After pruning the corresponding fuzzy subsets in the risk coefficient membership function according to the activation intensity, the centroid method is applied for defuzzification. The formula for the centroid method is provided below:

[0105] (7)

[0106] In equation (7): f(x) is the membership function of the risk coefficient, and x is the value of the risk coefficient in the domain of discourse.

[0107] The core idea of ​​the centroid method is to achieve precise output by calculating the centroid position of the fuzzy number. This is represented by dividing the first moment and zeroth moment of the membership function of the fuzzy subset within the universe of discourse. The fuzzy subsets output by the above fuzzy inference are "ZO, S". By taking the union of the membership functions of the two fuzzy subsets and calculating the centroid, the risk coefficient of this module can be output. The fuzzy inference simulation results are as follows... Figure 4 As shown.

[0108] Step 009: Divide the precise risk coefficient in the [0,1] interval into five risk levels with an interval length of 0.2: "Safe" [0,0.2), "Caution" [0.2,0.4), "Caution" [0.4,0.6), "Warning" [0.6,0.8), and "Danger" [0.8,1], to provide a rollover warning for the vehicle.

[0109] To verify the superiority of the present invention, this embodiment uses a rollover time warning method based on LTR for comparison. The principle of rollover time warning is now provided as follows: Figure 5 As shown. In Figure 5 In this method, t1 represents the current time, and a represents the LTR value at the current time. This method linearly predicts the time required for the LTR to reach 1 by calculating the slope of the LTR between the current time and the previous time. The specific formula is as follows:

[0110] (8)

[0111] In equation (8): T is the rollover time.

[0112] In Simulink, a rollover time warning module was built, using a linear periodic function and a sine function as inputs. The saturation module was used to limit the TTR output to an upper limit of 1. The rollover time output effect is as follows. Figure 6 As shown. Figure 6 In China, the TTR early warning method can accurately predict rollover time when the LTR changes linearly. However, when the LTR changes nonlinearly, the rollover time output by this method may be unpredictably ahead or behind. This phenomenon is the root cause of false alarms and missed alarms.

[0113] To further compare the method of this invention with the TTR warning method, the steering conditions in Trucksim were set as hook and sine wave conditions, with vehicle speeds of 40 km / h and 60 km / h respectively, and a vehicle sprung mass of 4455 kg. The steering wheel angle and angular velocity under these conditions are as follows: Figure 7 As shown in the figure. A simulation comparison was conducted using the two early warning methods described above, and the early warning effects are as follows. Figure 7 As shown.

[0114] exist Figure 8 In (a), the TTR warning under the fishhook condition relies entirely on the LTR rate of change. At 2.5-5 seconds, the LTR reaches 0.8, at which point the vehicle is at its rollover limit. The rollover time given by the TTR is significantly longer than at 1-3 seconds and 5-6 seconds, which is unreasonable. Figure 8In (c), the warning method described in this invention provides strong warnings during the aforementioned key time periods, with an output risk coefficient of approximately 0.9. Taking 2.5-5 seconds as an example, the method provides the highest warning when the LTR reaches 0.4, providing sufficient advance warning and high accuracy. The warning intensity varies in a stepwise manner with LTR and steering wheel angle, avoiding false alarms caused by direct jumps. Under sinusoidal conditions, Figure 8 (b) The TTR warning effect shown is always 1 because this steering condition is a cyclic condition and the LTR change rate is not large. However, the LTR reaches 0.4 at 1.5, 4, 6.5 and 9 seconds. For commercial vehicles, an LTR of 0.4 is enough for the driver to clearly feel the vehicle roll, and the sensitivity of the TTR warning is insufficient. Figure 8 In (d), the method of the present invention can accurately and provide early warning at the aforementioned time points, and the warning intensity is approximately a fixed value outside the aforementioned time points, ensuring the stability of the warning. The simulation results above show that the method of the present invention can adapt to different steering conditions and can provide fast, accurate, and stable rollover warning for commercial vehicles.

[0115] To facilitate verification that the rollover warning method described in this invention can adapt to different loads, assuming that the steering wheel angle, vehicle speed, and LTR remain constant, and setting the sprung mass of the vehicle to a full load of 8455 kg, the warning effect of the method described in this invention under full load conditions is as follows: Figure 9 As shown. Comparison Figure 8 (c) Figure 8 (d) and Figure 9 (a) Figure 9 (b) It can be seen that when the vehicle is fully loaded, the warning method has higher warning intensity and warning gradient than when it is unloaded. Therefore, the rollover warning method of the present invention can adapt to different load changes.

Claims

1. A method for adaptive rollover warning of commercial vehicles under steering conditions based on fuzzy rules, characterized in that, The steps of this method are as follows: Step 1: Collect vehicle speed signal, steering wheel angular velocity and four-wheel vertical load, and calculate the load lateral transfer rate based on the four-wheel vertical load; Step 2: Calculate the membership degree of each fuzzy subset based on the membership functions of vehicle speed, steering wheel angular velocity, and load lateral transfer rate, and infer the fuzzy subset to which it belongs. In the vehicle speed membership function, fuzzy subsets corresponding to zero, small-medium, medium, medium-large, and large are set according to the range of variation of the vehicle speed signal; in the load lateral transfer rate membership function, fuzzy subsets corresponding to very small, small, medium, large, and very large are set within the universe of discourse of the absolute value of the load lateral transfer rate |LTR|; in the steering wheel angular velocity membership function, fuzzy subsets corresponding to very small, small, medium, large, and very large are set within the universe of discourse of the absolute value of the steering wheel angular velocity |dθ|. sw / dt| Defines fuzzy subsets corresponding to small, medium, and large within the domain of discourse; Step 3: Based on the fuzzy subsets assigned by vehicle speed, steering wheel angular velocity, and load lateral transfer rate, the membership degrees of the zero, low, medium, and high fuzzy subsets corresponding to the risk coefficients are obtained by fuzzy rule reasoning; the fuzzy rules are specified in the table below; Step 4: Obtain the accurate risk coefficient through defuzzification, and use it to provide early warning of vehicle rollover.

2. The adaptive rollover warning method for commercial vehicle steering conditions based on fuzzy rules according to claim 1, characterized in that, In step one, the equivalent vehicle speed is used as the input for the vehicle speed. The expression for the equivalent vehicle speed is as follows: （1） In equation (1), V xm For the equivalent vehicle speed, m h For cargo weight, v x This refers to the actual vehicle speed.

3. The adaptive rollover warning method for commercial vehicle steering conditions based on fuzzy rules according to claim 1, characterized in that, The formula for calculating the lateral load transfer rate based on the vertical load of the four wheels in step one is as follows: （2） In equation (2): F z11 F is the vertical force on the left front wheel. z21 F is the vertical force on the left rear wheel. z12 F is the vertical force on the right front wheel. z22 This is the vertical force on the right rear wheel.

4. The adaptive rollover warning method for commercial vehicle steering conditions based on fuzzy rules according to claim 1, characterized in that, In step two, the membership function is represented using a trapezoidal rule; the standard form of the trapezoidal rule used is as follows: （3） In equation (3): m is the membership degree, a is the x-coordinate of the first endpoint, b is the x-coordinate of the second endpoint, c is the x-coordinate of the third endpoint, and d is the x-coordinate of the fourth endpoint; If equation (3) is rewritten when the x-coordinates of the first and second endpoints coincide or the x-coordinates of the third and fourth endpoints coincide, the following is true: （4） In equation (4), m1 is the membership degree when the horizontal coordinates of the first and second endpoints coincide, and m2 is the membership degree when the horizontal coordinates of the third and fourth endpoints coincide.

5. The adaptive rollover warning method for commercial vehicle steering conditions based on fuzzy rules according to claim 4, characterized in that, In step two, the endpoint coordinates of the vehicle speed membership function （5） Load lateral transfer rate membership function endpoint coordinates （6） Steering wheel angular velocity membership function endpoint coordinates （7） Risk coefficient membership endpoint coordinates （8）。 6. The adaptive rollover warning method for commercial vehicle steering conditions based on fuzzy rules according to claim 1, characterized in that, In step four, the precise risk coefficient in the range [0,1] is divided into five risk levels with an interval length of 0.2: "Safe" [0,0.2), "Caution" [0.2,0.4), "Caution" [0.4,0.6), "Warning" [0.6,0.8), and "Danger" [0.8,1], in order to provide a warning for vehicle rollover.

Images