CN115071441A - Extendible control torque constraint method and system based on particle swarm - Google Patents

Abstract

The invention relates to a particle swarm extension-based torque constraint control method and a particle swarm extension-based torque constraint control system. The method comprises the steps of obtaining a slip rate value s of a wheel and an actual barycenter slip angle beta of an automobile as characteristic quantities, and establishing a coordinate system of an extension control domain by taking the slip rate value s as an abscissa and the actual barycenter slip angle beta as an ordinate; and carrying out regional division on the extensible control domain: acquiring the linear region boundary of the actual barycenter slip angle gain of the automobile as the boundary beta of the vertical coordinate of the classical domain ₁ Obtaining a slip threshold s of the wheel _max As the abscissa boundary s of the classical domain ₁ (ii) a Calculating the slip rate optimal value s of the wheel by the particle swarm algorithm according to the slip rate value s of the current state of the wheel and the actual mass center slip angle beta of the current state of the automobile ₂ And corresponding optimal centroid sideDeclination angle beta ₂ (ii) a Judging the domain of the characteristic quantity and implementing a corresponding constraint mode; the invention realizes the self-adaptive vehicle wheel torque constraint mode along with the change of the road working condition.

Description

Extendible control torque constraint method and system based on particle swarm

Technical Field

The invention relates to the technical field of electric automobiles, in particular to a particle swarm extension-based torque control constraint method and a particle swarm extension-based torque control constraint system.

Background

The distributed hub motor driven automobile has the advantages that the space in the hub is small, the motor is prone to generating faults, accordingly, the driving performance of the distributed hub motor driven automobile is affected, and the driving stability of the distributed hub motor driven automobile is affected in severe cases. Because the road surface condition that the car travels is changeable, the car is easy to skid when meeting the low-adhesion road surface condition, the torque of the wheel needs to be adjusted in order to prevent the wheel from skidding, but the torque is not restricted, the situation of the wheel skidding is easy to be promoted, and further the stability is deteriorated.

Disclosure of Invention

Therefore, it is necessary to provide a particle swarm-based extension control torque constraint method and a particle swarm-based extension control torque constraint system for solving the problem that an extension domain boundary cannot be adjusted according to the driving condition of an automobile.

In order to achieve the purpose, the invention adopts the following technical scheme:

the torque constraint method based on particle swarm expandable control comprises the following steps:

s1, obtaining a slip rate value s of the wheel and an actual mass center slip angle beta of the automobile as characteristic quantities to form Q (s, beta), and establishing a coordinate system of an extensible control domain by taking the slip rate value s as a horizontal coordinate and the actual mass center slip angle beta as a vertical coordinate;

s2, carrying out regional division on the extension control domain:

classical domain: ((-s) ₁ ,s ₁ )，(-β ₁ ,β ₁ ))；

And (3) extension domain: ((-s) ¹ ₁ ,s ¹ ₁ )，(-β ¹ ₁ ,β ¹ ₁ ) In which-s) ¹ ₁ ∈(-s ₂ ,-s ₁ )，s ¹ ₁ ∈(s ₁ ,s ₂ )，-β ¹ ₁ ∈(-β ₂ ,-β ₁ )，β ¹ ₁ ∈(β ₁ ,β ₂ )；

Non-domain: ((-s) ² ₁ ,s ² ₁ )，(-β ² ₁ ,β ² ₁ ) In which-s) ² ₁ ∈(-∞,-s ₂ )，s ² ₁ ∈(s ₂ ,∞)，-β ² ₁ ∈(-∞,-β ₂ )，β ² ₁ ∈(β ₂ , ∞)；

S3, acquiring a linear region boundary of the actual mass center slip angle gain of the automobile as a boundary beta of a vertical coordinate of the classical domain ₁ Obtaining a slip rate threshold s of the wheel _max As the abscissa boundary s of the classical domain ₁ I.e. s ₁ ＝s _max ；

S4, calculating the slip rate optimal value s of the wheel by the particle swarm algorithm according to the slip rate value s of the current state of the wheel and the actual mass center slip angle beta of the current state of the automobile ₂ And the slip rate optimum s ₂ Corresponding optimal centroid slip angle beta ₂ ；

S5, judging whether the characteristic quantity is in a classical domain, if so, implementing torque constraint of the classical domain on the wheels of the automobile;

s6, when the characteristic quantity is not in the classical domain, judging whether the characteristic quantity is in an extension domain, if so, implementing torque constraint of the extension domain on the wheels of the automobile;

s7, when the characteristic quantity is not in an extension domain, the characteristic quantity is in a non-domain, and the non-domain torque constraint is implemented on the wheels of the automobile;

the optimum value of slip ratios ₂ And its corresponding optimum centroid slip angle beta ₂ The calculation method of (2) is as follows:

(1) obtaining an actual value ω of yaw rate of the wheels _z According to the actual value omega of the yaw angular velocity _z And setting a fitness function according to the actual centroid slip angle beta

Wherein v is _x Is the vehicle speed of the wheel, μ is the road adhesion coefficient, g is the gravitational acceleration;

(2) taking one wheel of the automobile as a target individual and all wheels of the automobile as a population, and taking the actual value omega of the yaw rate of each wheel _z Forming particles by taking the corresponding actual centroid side deflection angle beta as a particle parameter, and forming particle groups by all the particles;

(3) acquiring particle parameters of the population in a historical optimal state under an actual road adhesion coefficient to serve as a historical population optimal solution, and acquiring particle parameters of the target individual in the historical optimal state under the actual road adhesion coefficient to serve as a historical individual optimal solution;

(4) calculating an adaptive value of the target particle by taking the actual particle parameter of the target individual as the target particle through the fitness function;

(5) comparing the adaptive value of the target particle with the adaptive value of the historical individual optimal solution, if the adaptive value of the target particle is greater than the adaptive value of the historical individual optimal solution, replacing the particle parameter of the target particle with the particle parameter of the historical individual optimal solution, updating the individual optimal solution, and if the adaptive value of the target particle is less than the adaptive value of the historical individual optimal solution, keeping the particle parameter of the historical individual optimal solution inconvenient;

(6) comparing the adaptive value of the target particle with the adaptive value of the historical population optimal solution, if the adaptive value of the target particle is greater than the adaptive value of the historical population optimal solution, replacing the particle parameter of the target particle with the particle parameter of the historical population optimal solution, updating the population optimal solution, and if the adaptive value of the target particle is less than the adaptive value of the historical population optimal solution, keeping the particle parameter of the historical population optimal solution inconvenient;

(7) iteratively updating the particle parameters of the target particles according to the individual optimal solution and the new population optimal solution, judging whether the iteration times reach a preset maximum iteration number, if not, returning the updated target particles to the step (3), if so, stopping iteration, and calculating the optimal value s of the slip rate according to the actual value of the yaw angular velocity after the iteration of the target particles is finished through a slip rate calculation formula ₂ Taking the actual centroid slip angle after the target particle iteration is finished as the optimal centroid slip angle beta ₂ 。

Further, the iterative update method of the target particle comprises the following steps:

obtaining a maximum yaw rate of the target particles

The inertia weight tau of the target particle, the random number rand () of the target particle, which is more than 0 and less than 1, and the population optimal solution p of the kth generation of the target particle _g Individual optimal solution p of kth generation of the target particle _i Acceleration constant c of the target particle ₁ And an acceleration constant c of the target particle ₂ ；

Actual value of yaw rate according to k generation of target particle

And centroid slip angle of kth generation of the target particle

Calculating the k +1 th generation actual value of the yaw rate of the target particle

Calculating the centroid slip angle of the k +1 th generation of the target particle

Calculating the k-th generation yaw rate actual value of the target particle

And centroid slip angle of kth generation of the target particle

Updating to the k +1 th generation actual value of the yaw rate of the target particle

And centroid slip angle of the k +1 th generation of the target particle

Further, the inertia weight τ of the target particle is updated according to the change of the iteration number, and the calculation method of the inertia weight comprises the following steps:

obtaining a maximum weight τ of the target particle _max Minimum weight τ of the target particle _min ；

According to the iteration times t of the target particles _gen Calculating an inertial weight τ of the target particle:

further, the method for calculating the linear region boundary of the actual centroid slip angle gain of the automobile comprises the following steps:

obtaining the fitting parameters a required by the fitting calculation ₁₀ Fitting parameters b required for fitting calculation ₁₀ Fitting parameters c required for fitting calculation ₁₀ Fitting parameters d required for fitting calculation ₁₀ ；

According to the speed v of the automobile _x Calculating the front wheel steering angle linear zone boundary of the automobile by a fitting algorithm, namely the maximum value delta of the front wheel steering angle of the automobile _max ：

The maximum value delta of the front wheel steering angle of the automobile _max And calculating the linear region boundary of the actual barycenter sideslip angle gain of the automobile through an automobile two-degree-of-freedom model.

Further, the domain of the feature quantity is judged through a measure pattern, and the method for dividing the measure pattern comprises the following steps:

acquiring a correlation function K (S) of the position and the distance of the characteristic quantity in the extension control domain;

taking the measure mode of the characteristic quantity in the classical domain as M ₁ ：M ₁ ＝{S|K(S)＞1}；

Taking the measurement mode of the characteristic quantity in the extension domain as M ₂ ＝{S|0＜K(S)≤1}；

Taking the measurement mode of the characteristic quantity in the non-domain as M ₃ ＝{S|K(S)＜0}。

Further, the design method of the correlation function comprises the following steps:

obtaining an extension area interval X of the wheel _k And a classical domain interval X of said wheel _j ；

Connecting and extending the original point of the extension control domain and the characteristic quantity to obtain a connecting line, and intersecting the connecting line and the boundary of the extension domain to obtain an intersection point Q ₂ Intersection point Q of sums ₃ The above-mentionedIntersection point Q ₂ Has an abscissa of-s ₁ Said point of intersection Q ₃ Has an abscissa of s ₁ 。

Calculating the extension distance rho (Q, X) from the characteristic quantity to the extension domain _k )：

Calculating the correlation function k(s):

in one embodiment, when the characteristic quantity is in the measurement mode M ₁ When the maximum driving torque of the wheel is limited only by the peak torque of the in-wheel motor, regardless of the torque constraint limit of the wheel slip, i.e. the maximum driving torque of the wheel is limited only by the peak torque of the in-wheel motor

Wherein, P _max The rated power of the hub motor is shown, and n is the rotating speed of the motor.

In one embodiment, when the characteristic quantity is in the measurement mode M ₂ The value T is adjusted according to the driving torque in consideration of the wheel slip _e Calculating a maximum drive torque constraint of the wheel as T _max ＝[K(S)+1]T _e 。

In one embodiment, when the characteristic quantity is in the measurement mode M ₃ And when the wheel slip condition is serious, the actual mass center slip angle of the automobile enters a destabilizing state, and the maximum value of the wheel driving torque is constrained to be T _max ＝T _e 。

The invention also includes a particle swarm extension-based control torque constraint system for providing torque constraints for a plurality of wheels of an automobile, which adopts the particle swarm extension-based control torque constraint method as described above, the particle swarm extension-based control torque constraint system comprising:

the extension control domain coordinate system calculation module is used for acquiring a slip rate value s of the wheel and an actual centroid slip angle beta of the automobile as characteristic quantities to form Q (s, beta), and establishing a coordinate system of an extension control domain by taking the slip rate value s as a horizontal coordinate and the actual centroid slip angle beta as a vertical coordinate;

an extension control domain region division module, configured to perform region division on the extension control domain:

classical domain: ((-s) ₁ ,s ₁ )，(-β ₁ ,β ₁ ))；

And (3) extension domain: ((-s) ¹ ₁ ,s ¹ ₁ )，(-β ¹ ₁ ,β ¹ ₁ ) In which-s) ¹ ₁ ∈(-s ₂ ,-s ₁ )，s ¹ ₁ ∈(s ₁ ,s ₂ )，-β ¹ ₁ ∈(-β ₂ ,-β ₁ )，β ¹ ₁ ∈(β ₁ ,β ₂ )；

Non-domain: ((-s) ² ₁ ,s ² ₁ )，(-β ² ₁ ,β ² ₁ ) In which-s) ² ₁ ∈(-∞,-s ₂ )，s ² ₁ ∈(s ₂ ,∞)，-β ² ₁ ∈(-∞,-β ₂ )，β ² ₁ ∈(β ₂ , ∞)；

A classical domain boundary calculation module for obtaining a linear region boundary of the actual centroid slip angle gain of the automobile as a boundary beta of a vertical coordinate of the classical domain ₁ Obtaining a slip rate threshold s of the wheel _max As the abscissa boundary s of the classical domain ₁ I.e. s ₁ ＝s _max ；

An extension control domain boundary calculation module, configured to calculate an optimal slip ratio value s of the wheel by a particle swarm algorithm using the slip ratio value s of the current state of the wheel and the actual centroid slip angle β of the current state of the automobile ₂ And the slip rate optimum s ₂ Corresponding optimal centroid slip angle beta ₂ ；

The characteristic quantity domain judging module is used for judging whether the characteristic quantity is in a classical domain or not, and if so, implementing the torque constraint of the classical domain on the wheels of the automobile; when the characteristic quantity is not in the classical domain, judging whether the characteristic quantity is in an extension domain, if so, implementing torque constraint of the extension domain on the wheels of the automobile; when the characteristic quantity is not in the extension area, the characteristic quantity is in the non-area, and the non-area torque constraint is carried out on the wheels of the automobile.

The technical scheme provided by the invention has the following beneficial effects:

the method adopts an extension control theory to divide extension control domains respectively, realizes a self-adaptive vehicle wheel torque constraint mode along with the change of the road surface working condition, leads the vehicle wheel torque output constraint to change along with the change of the vehicle wheel torque output constraint under different working conditions, and searches the optimal extension domain boundary of the extension control domains for the road surfaces under different attachment conditions through particle groups.

Drawings

FIG. 1 is a flow chart of a particle swarm-based expandable control torque constraint method of the present invention;

FIG. 2 is a schematic diagram of a torque coordinated distribution system based on FIG. 1;

FIG. 3 is a three-dimensional simulation of the fuzzy rule of FIG. 1;

FIG. 4 is a schematic diagram of the structure of an extensible control domain based on the extensible control torque constraint of FIG. 2;

fig. 5 is a schematic structural diagram of a stable region of the yaw moment control system based on fig. 2;

FIG. 6 is a simulation of the torque distribution system based on FIG. 2 taking into account torque losses;

FIG. 7 is a simulation diagram of the single-wheel torque loss system based on FIG. 2;

FIG. 8 is a simulation diagram of the two-wheel torque loss system based on FIG. 2;

FIG. 9 is a logic diagram based on the torque coordinated distribution system of FIG. 2.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The particle swarm extension-based torque constraint control method solves the technical problem that the extension domain boundary can not be adjusted according to the driving condition of an automobile in the prior art. The invention adopts the extension control theory to divide extension control domains respectively, realizes the self-adaptive vehicle wheel torque constraint mode along with the change of the road surface working condition, leads the vehicle wheel torque output constraint to change along with the change of the vehicle wheel torque output constraint under different working conditions, and searches the optimal extension domain boundary of the extension control domains for the road surfaces under different attachment conditions through particle groups.

As shown in fig. 1, in the particle swarm optimization-based expandable control torque constraint method of the embodiment, the optimal expandable domain boundary is obtained by using a particle swarm optimization, so as to determine the expandable control domain, and the wheel torque constraint mode is determined by using the expandable control domain.

The constraint method of the invention comprises the following steps:

s1, obtaining a slip rate value s of the wheel and an actual mass center slip angle beta of the automobile as characteristic quantities to form Q (s, beta), and establishing a coordinate system of an extensible control domain by taking the slip rate value s as a horizontal coordinate and the actual mass center slip angle beta as a vertical coordinate;

s2, carrying out regional division on the extension control domain:

classical domain: ((-s) ₁ ,s ₁ )，(-β ₁ ,β ₁ ))；

And (3) extension domain: ((-s) ¹ ₁ ,s ¹ ₁ )，(-β ¹ ₁ ,β ¹ ₁ ) In which-s) ¹ ₁ ∈(-s ₂ ,-s ₁ )，s ¹ ₁ ∈(s ₁ ,s ₂ )，-β ¹ ₁ ∈(-β ₂ ,-β ₁ )，β ¹ ₁ ∈(β ₁ ,β ₂ )；

Non-domain: ((-s) ² ₁ ,s ² ₁ )，(-β ² ₁ ,β ² ₁ ) In which-s) ² ₁ ∈(-∞,-s ₂ )，s ² ₁ ∈(s ₂ ,∞)，-β ² ₁ ∈(-∞,-β ₂ )，β ² ₁ ∈(β ₂ , ∞)；

S3, acquiring a linear region boundary of the actual mass center slip angle gain of the automobile as a boundary beta of a vertical coordinate of the classical domain ₁ Obtaining a slip rate threshold s of the wheel _max As the abscissa boundary s of the classical domain ₁ I.e. s ₁ ＝s _max ；

S4, calculating the slip rate optimal value s of the wheel by the particle swarm algorithm according to the slip rate value s of the current state of the wheel and the actual mass center slip angle beta of the current state of the automobile ₂ And the slip rate optimum s ₂ Corresponding optimal centroid slip angle beta ₂ ；

S5, judging whether the characteristic quantity is in a classical domain, if so, implementing torque constraint of the classical domain on the wheels of the automobile;

s6, when the characteristic quantity is not in the classical domain, judging whether the characteristic quantity is in an extension domain, if so, implementing torque constraint of the extension domain on the wheels of the automobile;

and S7, when the characteristic quantity is not in the extension domain, the characteristic quantity is in a non-domain, and the non-domain torque constraint is implemented on the wheels of the automobile.

The particle swarm extension-based control torque constraint system comprises an extension control domain coordinate system calculation module, wherein the extension control domain coordinate system calculation module is used for acquiring a slip rate value s of the wheel and an actual centroid slip angle beta of the automobile as characteristic quantities to form Q (s, beta), and establishing a coordinate system of an extension control domain by taking the slip rate value s as an abscissa and the actual centroid slip angle beta as an ordinate; an extension control domain region division module, configured to perform region division on the extension control domain: classical domain: ((-s) ₁ ,s ₁ )， (-β ₁ ,β ₁ ) ); an extensible domain: ((-s) ¹ ₁ ,s ¹ ₁ )，(-β ¹ ₁ ,β ¹ ₁ ) In which-s) ¹ ₁ ∈(-s ₂ ,-s ₁ )，s ¹ ₁ ∈(s ₁ ,s ₂ )，-β ¹ ₁ ∈(-β ₂ ,-β ₁ )，β ¹ ₁ ∈(β ₁ ,β ₂ ) (ii) a Non-domain: ((-s) ² ₁ ,s ² ₁ )，(-β ² ₁ ,β ² ₁ ) In which-s) ² ₁ ∈(-∞,-s ₂ )，s ² ₁ ∈(s ₂ ,∞)，-β ² ₁ ∈(-∞,-β ₂ )，β ² ₁ ∈(β ₂ Infinity); a classical domain boundary calculation module for obtaining a linear region boundary of the actual centroid slip angle gain of the automobile as a boundary beta of a vertical coordinate of the classical domain ₁ Obtaining a slip rate threshold s of the wheel _max As the abscissa boundary s of the classical domain ₁ I.e. s ₁ ＝s _max (ii) a An extension control domain boundary calculation module, configured to calculate an optimal slip ratio value s of the wheel by a particle swarm algorithm using the slip ratio value s of the current state of the wheel and the actual centroid slip angle β of the current state of the automobile ₂ And the slip rate optimum s ₂ Corresponding optimal centroid slip angle beta ₂ (ii) a The characteristic quantity domain judging module is used for judging whether the characteristic quantity is in a classical domain or not, and if so, implementing the torque constraint of the classical domain on the wheels of the automobile; when the characteristic quantity is not in the classical domain, judging whether the characteristic quantity is in an extension domain, if so, implementing torque constraint of the extension domain on the wheels of the automobile; when the characteristic quantity is not in the extension area, the characteristic quantity is in the non-area, and the non-area torque constraint is carried out on the wheels of the automobile.

The extension control torque restraint system based on the particle swarm is contained in a torque coordination distribution system, which is specifically explained below.

As shown in fig. 2, a torque coordination distribution system is designed for the condition that the torque loss of a hub motor of a distributed drive automobile under a low-adhesion road surface seriously affects the operation stability and the driving safety of the automobile, and mainly comprises a nonlinear error feedback torque control system, a particle swarm extension control torque constraint system, an automobile stability-based yaw moment control system and a torque distribution system based on torque loss. When the nonlinear error feedback torque control system works, the nonlinear error feedback torque control method is adopted.

The following is a detailed description of the non-linear error feedback torque control system:

as the road condition may change all the time, in order to prevent the tires of the automobile from skidding and increase the applicable working condition range of the stability control strategy, a fuzzy control algorithm is adopted to adaptively adjust the wheel slip rate threshold value.

By the deviation delta omega of the actual value of the yaw rate of the vehicle from the reference value _z (Δω _z ＝ω _z -ω _z ') and a slip rate value s for the current wheel, a slip rate threshold value s _max As an output, fuzzy rule setting of the slip rate threshold value is carried out, for example, when the wheel slip rate value s and the yaw rate are larger, the vehicle is in urgent need of reducing wheel slip to ensure the stability of the vehicle, and the wheel slip rate threshold value output by fuzzy control is smaller. The fuzzy logic rule table established according to the multiple simulation tests is shown in the following table.

Fuzzy rule table for slip rate threshold value

The fuzzy rule three-dimensional effect graph obtained by the fuzzy rule and the corresponding input and output membership function can judge and output the slip rate threshold value adaptive to the current working condition according to the parameter estimation condition as shown in fig. 3, and further carry out wheel anti-slip control to restrict the wheel driving torque.

Based on the slip threshold value of the fuzzy control, a non-linear error feedback torque control system is designed to obtain a desired torque to prevent the wheel from slipping. The nonlinear error feedback control principle is to make nonlinear combination of errors and then output the torque control demand, as shown in fig. 1, wherein the input of the nonlinear negative feedback control is the deviation e between the wheel slip ratio estimated value and the slip ratio threshold value ₁ Deviation e between actual value of wheel drive torque and expected value of torque ₂ Calculating the driving torque adjusting value T by combining the errors of the two through nonlinearity _e To prevent the wheel from slipping. Beta is a centroid slip angle, and the designed nonlinear error feedback control mathematical expression is as follows:

wherein u is ₀ For the nonlinear control law, k ₀ Is a gain coefficient, c ₀ Is a damping coefficient, h ₀ For the sampling time, a factor b is compensated ₀ ＝1/J _ω (inertia of wheel 0.9kg · m2), coefficient of regulation r ₀ ＝0.05/h ₀ ，r ₀ The function of (a) is to adjust the control force of the nonlinear feedback control system, the value can be determined by empirical value or multiple tests, fst (·) is the synthesized function of the fastest control, and

the following describes a torque constraint system based on particle swarm extension control specifically:

designing torque constraint control by considering wheel slip prevention, wherein the wheel torque output constraint requirements under different working conditions are different, so that extension control domains are respectively divided by adopting an extension control theory, and an adaptive vehicle wheel torque constraint mode along with the change of the road working conditions is realized; the extension control domain is divided as shown in fig. 4, and the intersection points of the connecting line of the original point and the feature quantity in the extension set and the boundary of the extension domain and the classical domain from left to right are Q ₁ 、Q ₂ 、Q ₃ 、Q ₄ And the slip rate of the automobile wheels in the classical domain is small, the mass center slip angle of the automobile is small, the automobile is in a working condition with good road adhesion condition, and the torque restriction of wheel slip is not needed at the moment. The slip rate and the centroid slip angle of the automobile wheel in the extension domain tend to increase, at the moment, constraint limitation needs to be applied to torque distribution, the state is kept to be the best in the extension domain, and the torque is prevented from continuously increasing to enter the non-domain. The slip ratio and the mass center slip angle of the wheels in the non-domain are large, the automobile is about to be unstable or enters an unstable state, the torque of the wheels needs to be strictly restricted at the moment, and the safety and the stability of the automobile are ensured as far as possible.

The method comprises the following steps that an extension set is required to be divided in an extension control domain, the boundary of each region is determined, and then the wheel torque constraint mode of each control region is determined, wherein the design steps are as follows:

(1) selecting characteristic quantities

The characteristic quantities are used to indicate a wheel slip condition and a vehicle steady state. Therefore, in order to distinguish the wheel slip and driving stability boundaries and regions in the classical domain, the extension domain and the non-domain, the slip value s of the wheel and the actual centroid slip angle β of the vehicle are selected as characteristic quantities, and the two values form Q (s, β).

(2) Partitioning of scalable sets

The method comprises the following steps of firstly dividing a classical domain boundary, wherein the classical domain boundary is relatively easy to divide, and a linear region boundary of a yaw velocity gain is used as an actual centroid yaw angle classical domain boundary. Calculating the maximum value of the actual centroid slip angle of the linear region through a fitting relation, wherein the value of the actual centroid slip angle is the classical domain boundary beta ₁ . The empirical formula of the front wheel steering angle limit value and the vehicle speed is as follows:

wherein, a ₁₀ ，b ₁₀ ，c ₁₀ And d ₁₀ For the fitting parameters, 0.05, 0.07, 0.6 and 13.3, respectively, which can be determined empirically or by multiple experiments; calculating the maximum value delta of the front wheel rotation angle in the linear region _max Substituted into vehicle twoThe classical domain boundary beta can be calculated by a freedom degree model ₁ 。

Classical domain boundary s of wheel slip ₁ Bounded by a slip threshold value of the fuzzy control output, i.e. s ₁ ＝s _max . And when the wheel slip rate is smaller than the slip rate threshold value, the wheel slip state is considered to be stable and is in the classical domain.

And secondly, dividing the extension domain boundary. The particle swarm algorithm does not depend on strict mathematics of the optimization problem and accurate mathematical description of a target function and a constraint condition, only a corresponding evaluation function needs to be designed, and the extension domain boundary is a two-dimensional boundary with small dimension, so that the particle swarm algorithm is adopted for dividing the extension domain boundary to iteratively search the optimal boundary. In each iteration of the algorithm, the particles continuously solve the optimal solution p according to the individual _i And the population optimal solution p so far _g Both extrema update themselves. Further, the particle velocity and position are updated according to the following formula:

wherein the content of the first and second substances,

is the velocity and position of the particle in the k-th generation,

is the maximum velocity of the particle, whose value is too large to fly through the optimal solution, rand () is a random number greater than 0 and less than 1, and the learning factor is c ₁ ＝c ₂ τ is an inertial weight, which is used to balance the local and global optima, givenThe value of tau is shown as the following formula:

wherein the maximum weight τ _max 1.4, minimum weight τ _min ＝0.4，t _gen Is the number of iterations.

The fitness function is set as:

therefore, the specific process of determining the extension domain boundary by the particle swarm algorithm is as follows:

step 1, obtaining the actual yaw velocity value omega of the wheels _z Based on the actual value ω of yaw rate _z Setting fitness function with actual centroid slip angle beta

Wherein v is _x Is the vehicle speed of the wheel, μ is the road adhesion coefficient, g is the gravitational acceleration;

step 2, taking one wheel of the automobile as a target individual and all wheels of the automobile as a population, and taking the actual value omega of the yaw rate of each wheel _z Forming particles by taking the corresponding actual centroid side deflection angle beta as a particle parameter, and forming particle groups by all the particles;

step 3, acquiring particle parameters of the population in the historical optimal state under the actual road adhesion coefficient to serve as the optimal solution of the historical population, and acquiring particle parameters of the target individual in the historical optimal state under the actual road adhesion coefficient to serve as the optimal solution of the historical individual;

step 4, calculating the adaptive value of the target particle by the target particle through a fitness function by taking the actual particle parameter of the target individual as the target particle;

step 5, comparing the adaptive value of the target particle with the adaptive value of the historical individual optimal solution, if the adaptive value of the target particle is greater than the adaptive value of the historical individual optimal solution, replacing the particle parameter of the historical individual optimal solution with the particle parameter of the target particle, updating the individual optimal solution, and if the adaptive value of the target particle is less than the adaptive value of the historical individual optimal solution, keeping the particle parameter of the historical individual optimal solution inconvenient;

step 6, comparing the adaptive value of the target particle with the adaptive value of the historical population optimal solution, if the adaptive value of the target particle is greater than the adaptive value of the historical population optimal solution, replacing the particle parameter of the historical population optimal solution with the particle parameter of the target particle, updating the population optimal solution, and if the adaptive value of the target particle is less than the adaptive value of the historical population optimal solution, keeping the particle parameter of the historical population optimal solution inconvenient;

and 7, carrying out iterative updating on the particle parameters of the target particles according to the individual optimal solution and the new population optimal solution, judging whether the iteration times reach a preset maximum iteration number, if not, returning the updated target particles to the step 3, if so, stopping the iteration, and calculating the optimal value s of the slip rate according to the actual value of the yaw angular velocity after the iteration of the target particles through a slip rate calculation formula ₂ Taking the actual centroid slip angle after the iteration of the target particles is finished as the optimal centroid slip angle beta ₂ 。

(3) Design of relevance function

Designing a correlation function according to the position and the distance of the characteristic quantity in the control domain, wherein the extension distance from the characteristic point to the extension domain is as follows:

wherein, X _k If the extension range is, the correlation function k(s) is:

wherein X _j The domain is a classical domain interval, so that the domain where the characteristic quantity is located can be judged more conveniently through the calculation result of the correlation function K (S).

(4) Measure pattern partitioning

The measure mode of the characteristic quantity, namely the measure mode M, can be judged by the correlation function K (S) ₁ When { S | k (S) > 1} the feature is in the classical domain; measure mode M ₂ When the feature quantity is in an extension domain, the { S |0 is less than K (S) is less than or equal to 1 }; measure mode M ₃ When { S | k (S) < 0}, the feature quantity is in the non-domain.

Thus at M ₁ In the measure mode, the torque constraint limitation of wheel slip is not required to be considered, and the torque is only constrained by the peak torque of the hub motor, namely

P _max The rated power of the hub motor is shown, and n is the rotating speed of the motor.

At M ₂ In the measure mode, a certain degree of wheel slip needs to be considered. Driving torque regulating value T output by feedback torque control system according to non-linear error _e Design torque maximum constraint T _max Comprises the following steps: t is _max ＝[K(S)+1]T _e 。

At M ₃ In a measure mode, the wheel slip condition is serious, the mass center slip angle also enters a destabilization state, and the maximum value of the design torque is restricted by T _max Comprises the following steps: t is _max ＝T _e 。

The constraint mode of the wheel output torque is judged through the extension domain division of the wheel slip rate value and the actual centroid slip angle, and the optimal extension domain boundary is searched for the road surfaces under different attachment conditions through the particle swarm optimization.

The following is a detailed description of a yaw moment control system based on vehicle stability:

the loss of the output torque of the hub motor can cause sudden change of the yaw moment of the automobile. An excessive yaw moment is generated in the process, so that the automobile can possibly change the course angle and the driving track rapidly and enter an unstable working condition, and further the driving danger is increased. Therefore, the adaptive yaw moment constraint should be available for the optimal control of the yaw motion of the automobile and the control strategy of the distribution of the driving torque, so as to ensure the running stability of the automobile.

Firstly, a control boundary is determined according to the automobile state, namely a stability control criterion is determined, the stability criterion of the automobile determines whether a yaw moment control system is started or not, and an actual mass center slip angle-an actual mass center slip angular velocity (beta-v) is adopted _β ) The phase plane is used as the criterion of the stability of the automobile.

The method comprises dividing the stable region of the automobile by bilinear method and limit ring method, i.e. adding an ellipse tangent to two straight lines in the region divided by bilinear method, and using the ellipse as the stable region, as shown in FIG. 5, with straight line l ₁ And l ₂ Phase plane stability boundaries determined for bilinear methods, while ellipses and/ ₁ And l ₂ The tangency is realized, a finally determined stable domain is formed in the tangent, so that two tangent points are formed by the simultaneous connection of two straight lines and an ellipse equation, and the major and minor semiaxes a and b of the ellipse are finally calculated:

wherein, c ₁₁ And c ₂₂ According to the actual phase diagram, the automobile stability criterion is obtained by the method, namely if the automobile state is outside an elliptical stable area, the yaw moment control system is required to act and output an additional yaw moment for recovering the stability, so that the torque distribution layer distributes the torque to enable the automobile to recover the stability. The lengths of the semiaxes of the stable area ellipses under different road adhesion coefficients obtained by the method are shown in the following table:

stable boundary parameter table

The fitting formula for obtaining the change of the stable boundary ellipse semi-axis along with the road adhesion coefficient according to the parameters in the table is as follows:

and (3) calculating the additional yaw moment required by the stability of the automobile by adopting a sliding mode control algorithm, wherein the sliding mode control law and the sliding mode surface can be changed according to the characteristic requirements of the controlled system and are unrelated to the disturbance and other parameters of the system.

Firstly, calculate the carReferring to the yaw moment, as the road conditions are various and the road surface adhesion coefficient is also changed continuously, the automobile theory can know that the automobile yaw angular speed and the centroid slip angle are constrained by the adhesion conditions, and the two are constrained as follows according to the adhesion conditions:

then the ideal centroid slip angle beta of the automobile _d And yaw angular velocity ω _d Respectively as follows:

wherein

Is the coefficient of stability, v, of the vehicle _x Is the speed of the vehicle, delta is the angle of rotation of the front wheel, l _r Is the distance from the center of mass to the rear axis,/ _f Is the distance from the center of mass to the front axle of the automobile, L is the wheel base of the wheels of the automobile, k ₂ Is the rear wheel side yaw stiffness k of the automobile ₂ ，k ₁ Is the front wheel side deflection rigidity of the automobile, and m is the whole automobile mass

The designed sliding mode function is as follows: u ═ ω (ω) _z -ω _d )+h ₂ (β-β _d )，

And (3) carrying out derivation on the sliding mode control rate u, wherein the derivation formula is as follows:

the approach law of the sliding mode controller is as follows:

wherein beta is the actual centroid slip angle of the automobile, epsilon is the torque loss coefficient of the automobile, and h ₁ Calculated by sliding-mode control rateWeighting factor, h ₂ Is a weighting coefficient h calculated by a sliding mode control approach rate ₂ 。

Because the sliding mode control has the buffeting problem, a filtering method can be adopted, a low-pass filter is used for a switching function of the sliding mode control, a smooth signal is obtained, and the buffeting of the system is reduced. And finally obtaining an additional yaw moment for maintaining the stable running requirement of the automobile by a simultaneous three-degree-of-freedom model of the automobile body as follows:

where Fyfr is the front right tire lateral force, Fyrr is the rear right tire lateral force, Fyfl is the front left tire lateral force, F _yrl Is the rear left tire lateral force.

The torque distribution system based on torque loss is described in detail below:

in order to ensure the yaw stability of the automobile, the torque distribution is carried out on the four-wheel drive torque of the distributed drive automobile according to the additional yaw moment required by the stability of the automobile, which is obtained by a yaw moment control system based on the stability of the automobile. The four-wheel torque distribution adopts a quadratic programming algorithm. According to the theory of wheel dynamics, the tire force of the wheel is limited by the adhesion ellipse, the larger the longitudinal force of the wheel, the smaller the margin for the lateral force, and the lateral stability of the automobile during steering is affected. Therefore, the tire load rate is adopted to restrain the wheel torque, so that the lateral stability of the automobile is ensured, and the automobile is provided with four wheels as an example.

First, an optimization objective function is determined, and tire force is calculated according to a Dugoff tire model, and the relationship between the tire force and the tire adhesion force is as follows:

wherein, F _xij Is the longitudinal force of the tire, F _yij Is the tire lateral force, F _zij Is the vertical load of the tyre

The tire load rate is set as follows:

where i ∈ { f, r } denotes front and back, and j ∈ { l, r } denotes left and right.

The driving torque of the hub motor can be directly controlled, the longitudinal force and the lateral force of the tire are provided by the adhesive force, the transverse force is ensured to reduce the longitudinal force, and therefore, the longitudinal force is mainly used for setting a quadratic objective function as follows:

wherein E (rho) represents the average mean value of the tire load rates, the E (rho) is added to enable each tire load rate to be close, the utilization rate of each tire can be improved, and eta is a weighting coefficient used for coordinating the two-part optimization target ratio.

Constraints for optimal allocation are then determined, mainly the equality constraint of the additional yaw moment with the longitudinal driver model output driving force and the inequality constraint of the maximum torque constraint taking into account wheel slip, the constraints established being as follows:

wherein F _x For total driving force for vehicle running, P _max Is the rated power of the hub motor, n is the motor speed, s is the wheel slip ratio, T _m Actual value of drive torque, F _xfr Is the front right tire longitudinal force, F _xrr Is the rear right tire longitudinal force, F _xfl Is the front left tire longitudinal force, F _xrl Is the rear left tire longitudinal force.

And finally, determining an optimal distribution algorithm, optimally distributing the upper layer yaw moment by utilizing a quadratic programming theory, converting the target function into a standard quadratic form, wherein the quadratic programming function is as follows:

wherein x ═ F _xfl ,F _xfr ,F _xrl ,F _xrr ] ^T Expressing matrix transposition, and deducing a quadratic programming H matrix and a quadratic programming c matrix as follows by inequality constraint and a quadratic programming function:

F _zfl is the vertical load of the front left tire of the automobile, F _zfr Is the vertical load of the front and right tires of the automobile, F _zrl Is the rear left tire vertical load of the automobile, F _zrr The vertical load of the rear right tire of the automobile is obtained by an inequality constraint function, wherein A and b are respectively as follows:

b＝[T _flmax T _frmax T _rlmax T _rrmax ]，

where r radius of the wheel, T _flmax Is the maximum value of the driving torque, T, of the front left wheel of the automobile _frmax Is the maximum value of the driving torque, T, of the front right wheel of the automobile _rlmax Is the maximum value of the driving torque, T, of the rear left wheel of the automobile _rrmax Is the maximum value of the driving torque of the rear right wheel of the automobile, therefore, the constraint of the inequality Ax is less than or equal to b is an affine function, A _eq x＝b _eq The equality constraint is also an affine function, which is consistent with the characteristics of the convex quadratic programming problem, which is thus a convex quadratic programming problem.

In addition, since the four-wheel hub motor has torque loss, it is necessary to reconstruct torque distribution according to the degree of torque loss, where ∈ ═ epsilon _fl ,ε _fr ,ε _rl ,ε _rr ] ^T For quadratic programming function, the function is constrained according to inequality and epsilon is integrated to constraint coefficient matrix A _eq And b _eq In the middle, get

b _eq ＝[ΔM F _x ]

After the quadratic programming function is solved to obtain the optimal driving torque of the four wheels, the four-wheel driving torque distribution is further reconstructed according to the torque loss coefficient, and the output torque of the hub motor is controlled, so that the automobile can still have the capability of stable running after the motor torque is lost.

For the simulation analysis of the torque coordination distribution system, firstly, whether the torque distribution strategy considering the torque loss can ensure the driving stability of the automobile when the torque of the in-wheel motor is lost is preliminarily verified. And selecting a double-shift line working condition, setting the road adhesion coefficient to be 0.85, simulating to reduce the loss of the right front hub motor to 60% torque output, and setting the simulated initial vehicle speed and the target vehicle speed to be 60 km/h. The results are shown in FIG. 6. The Torque Distribution strategy in which the Torque loss is taken into account is denoted as tdtl (Torque Distribution fire Torque loss). As can be seen from fig. 6(a), the loss of the torque of the right front wheel causes the vehicle to have a yaw rate for driving to the right, and after the torque is redistributed to the four wheels, the yaw rate of the vehicle can follow the desired value and the vehicle can be safely driven, thereby improving the maneuverability of the vehicle. As can be seen from fig. 6(b), the centroid slip angle starts to increase after the torque loss of the right front wheel of the automobile, which seriously affects the stability of the automobile, and the torque distribution strategy considering the torque loss reconstructs the four-wheel torque, so that the error between the actual value of the centroid slip angle and the reference value is reduced, and the stability of the automobile is greatly increased. As can be seen from fig. 6(c), when the torque loss occurs in the right front wheel of the vehicle, the vehicle not only has poor steering stability, but also has a certain loss in vehicle speed, and the torque distribution strategy considering the torque loss can make the vehicle speed decrease relatively smaller in the double-lane action, thereby improving the dynamic performance of the vehicle to a certain extent. Therefore, the torque distribution strategy considering the torque loss can be analyzed and obtained, and the torque distribution strategy can play a role in ensuring the driving safety, the operation stability and certain dynamic property of the automobile on a road surface with good adhesion conditions.

When single-wheel torque loss simulation verification is carried out, a double-traverse-line working condition simulation test is selected, the torque loss of the right front wheel is reduced to 60%, the road adhesion coefficient is set to be 0.3, the initial vehicle speed and the target vehicle speed are both 60km/h, and the simulation result is shown in fig. 7. The torque distribution strategy in which the torque losses are taken into account is denoted TDTL. The wheel Slip constrained Torque coordinated Distribution strategy taking into account Torque losses is denoted as TDSRTL (Torque Distribution for Slip braking while Torque loss). As can be seen from the comparative simulation results of the two control strategies, in fig. 7(b), the torque coordinated split strategy (TDSRTL) that takes into account the wheel slip constraint can make the vehicle yaw rate response faster, and the vehicle drivability better, i.e., can respond faster to the driver's steering demand, than the torque split strategy (TDTL) that does not take into account the torque constraint. In the torque distribution strategy in FIG. 7(c), the centroid slip angle of the torque distribution strategy without considering the torque constraint at the time of 10.45s is-2.845 degrees, while the centroid slip angle of the torque coordination distribution control strategy with considering the wheel slip constraint is-2.446 degrees, so that 14.02 percent is reduced, the jitter of the centroid slip angle of the automobile can be eliminated, and the driving stability of the automobile is improved. As can be seen from the four-wheel torque output curve in fig. 7(d), after the torque of the right front wheel is lost, the drive torque of the left front wheel is reduced, and the drive torque of the right rear wheel is correspondingly increased according to the yaw moment control demand, but under the constraint control considering the wheel slip, the drive torque of the right rear wheel is reduced at any time in the process of increasing the torque so as to prevent the wheel slip, and the running safety of the automobile is ensured. In fig. 7(e), although the wheel torque is limited and the vehicle driving torque is reduced to some extent, the vehicle speed is reduced little and is reduced by about 0.5km/h only during the double lane shifting operation. In general, when the right front wheel has torque loss, the proposed coordinated torque distribution control strategy considering wheel slip redistributes the torque, ensures the yaw moment required by the stability of the automobile, and ensures the driving safety of the automobile.

And when the two-wheel torque loss simulation verification is carried out, the working condition that the two wheels generate the torque loss is designed, and the proposed torque coordination distribution control strategy is verified. And selecting a double-shift-line simulation test. The torque loss of the two wheels on the same side is set, the torque loss of the right front wheel is reduced to 80%, the torque loss of the right rear wheel is reduced to 50%, the road adhesion coefficient is 0.3, the initial vehicle speed and the target vehicle speed are both 60km/h, and the simulation result is shown in FIG. 8. From the simulation comparison result of the torque loss of the two wheels on the same side, it is seen that the torque distribution strategy (TDTL) in fig. 8(b) without considering the torque constraint generates a certain degree of sideslip when the lane is changed for the second time in the double lane shifting action, so that the automobile is in a dangerous driving state in the period of time, and the torque coordination distribution control strategy (TDSRTL) considering the wheel slip improves the sideslip phenomenon and ensures the driving safety of the automobile. As can be seen from fig. 8(c), although the torque distribution strategy without considering the torque constraint can reduce the centroid slip angle of the vehicle before 10s to ensure the stability of the vehicle, the centroid slip angle is increased instead to cause the wheel slip during the second lane change, and the torque coordination distribution control restrains the torque output to a certain extent under the extension judgment, but controls the wheel slip at the same time to ensure the stability of the vehicle as a whole. As can be seen from the four-wheel torque in fig. 8(d), since the torque loss occurs in both right two wheels, the yaw moment cannot be compensated by increasing the torque on the different-axis side as in the case of the single-wheel loss. The left wheel torque also decreases and follows the change in the right wheel torque while the left wheel fine tuning torque output maintains the yaw moment stable. As can be seen from FIG. 8(e), the lowest vehicle speed of the torque distribution strategy without considering the torque constraint is 56.5km/h, and the vehicle speed under the torque coordination distribution control can be kept between 57 km/h and 60km/h, so that the dynamic property of the vehicle is ensured to a certain extent.

The verification proves that the whole torque coordination distribution system can ensure that the automobile can stably and safely run when the motor torque is lost by distributing the torque of the four-wheel motor of the automobile.

The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. A particle swarm-based scalable control torque constraint method for providing torque constraints to a plurality of wheels of an automobile, the torque constraint method comprising the steps of:

s1, obtaining a slip rate value s of the wheel and an actual mass center slip angle beta of the automobile as characteristic quantities to form Q (s, beta), and establishing a coordinate system of an extensible control domain by taking the slip rate value s as a horizontal coordinate and the actual mass center slip angle beta as a vertical coordinate;

s2, carrying out region division on the extension control domain:

classical domain: ((-s) ₁ ,s ₁ )，(-β ₁ ,β ₁ ))；

And (3) extension domain: ((-s) ¹ ₁ ,s ¹ ₁ )，(-β ¹ ₁ ,β ¹ ₁ ) In which-s) ¹ ₁ ∈(-s ₂ ,-s ₁ )，s ¹ ₁ ∈(s ₁ ,s ₂ )，-β ¹ ₁ ∈(-β ₂ ,-β ₁ )，β ¹ ₁ ∈(β ₁ ,β ₂ )；

Non-domain: ((-s) ² ₁ ,s ² ₁ )，(-β ² ₁ ,β ² ₁ ) In which-s) ² ₁ ∈(-∞,-s ₂ )，s ² ₁ ∈(s ₂ ,∞)，-β ² ₁ ∈(-∞,-β ₂ )，β ² ₁ ∈(β ₂ ,∞)；

S3, acquiring the reality of the automobileBoundary of linear region of inter-centroid side slip angle gain as boundary beta of ordinate of the classical domain ₁ Obtaining a slip rate threshold s of the wheel _max As the abscissa boundary s of the classical domain ₁ I.e. s ₁ ＝s _max ；

S4, calculating the slip rate optimal value s of the wheel by the particle swarm algorithm according to the slip rate value s of the current state of the wheel and the actual mass center slip angle beta of the current state of the automobile ₂ And the slip rate optimum s ₂ Corresponding optimal centroid slip angle beta ₂ ；

S5, judging whether the characteristic quantity is in a classical domain, if so, implementing torque constraint of the classical domain on the wheels of the automobile;

s6, when the characteristic quantity is not in the classical domain, judging whether the characteristic quantity is in an extension domain, if so, implementing torque constraint of the extension domain on the wheels of the automobile;

s7, when the characteristic quantity is not in an extension domain, the characteristic quantity is in a non-domain, and the non-domain torque constraint is implemented on the wheels of the automobile;

the method is characterized in that:

the optimum value of slip rate s ₂ And its corresponding optimal centroid slip angle beta ₂ The calculation method of (2) is as follows:

(1) obtaining an actual value ω of yaw rate of the wheels _z According to the actual value omega of the yaw angular velocity _z And setting a fitness function according to the actual centroid slip angle beta

Wherein v is _x Is the vehicle speed of the wheel, μ is the road adhesion coefficient, g is the gravitational acceleration;

(2) taking one wheel of the automobile as a target individual and all wheels of the automobile as a population, and taking the actual value omega of the yaw rate of each wheel _z Forming particles by taking the corresponding actual centroid side deflection angle beta as a particle parameter, and forming particle groups by all the particles;

(3) acquiring particle parameters of the population in a historical optimal state under an actual road adhesion coefficient to serve as a historical population optimal solution, and acquiring particle parameters of the target individual in the historical optimal state under the actual road adhesion coefficient to serve as a historical individual optimal solution;

(4) calculating an adaptive value of the target particle by using the target particle through the fitness function by taking the actual particle parameter of the target individual as the target particle;

(5) comparing the adaptive value of the target particle with the adaptive value of the historical individual optimal solution, if the adaptive value of the target particle is greater than the adaptive value of the historical individual optimal solution, replacing the particle parameter of the target particle with the particle parameter of the historical individual optimal solution, updating the individual optimal solution, and if the adaptive value of the target particle is less than the adaptive value of the historical individual optimal solution, keeping the particle parameter of the historical individual optimal solution inconvenient;

(6) comparing the adaptive value of the target particle with the adaptive value of the historical population optimal solution, if the adaptive value of the target particle is greater than the adaptive value of the historical population optimal solution, replacing the particle parameter of the target particle with the particle parameter of the historical population optimal solution, updating the population optimal solution, and if the adaptive value of the target particle is less than the adaptive value of the historical population optimal solution, keeping the particle parameter of the historical population optimal solution inconvenient;

(7) iteratively updating the particle parameters of the target particles according to the individual optimal solution and the new population optimal solution, judging whether the iteration times reach a preset maximum iteration number, if not, returning the updated target particles to the step (3), if so, stopping iteration, and calculating the optimal value s of the slip rate according to the actual value of the yaw angular velocity after the iteration of the target particles is finished through a slip rate calculation formula ₂ Taking the actual centroid slip angle after the target particle iteration is finished as the optimal centroid slip angle beta ₂ 。

2. The particle swarm expandable control torque constraining method of claim 1, wherein the iterative update method of the target particle comprises the steps of:

obtaining a maximum yaw rate of the target particles

The inertia weight tau of the target particle, the random number rand () of the target particle, which is more than 0 and less than 1, and the population optimal solution p of the kth generation of the target particle _g Individual optimal solution p of kth generation of the target particle _i Acceleration constant c of the target particle ₁ And an acceleration constant c of the target particle ₂ ；

Actual value of yaw rate according to k generation of target particle

And centroid slip angle of kth generation of the target particle

Calculating the k +1 th generation actual value of the yaw rate of the target particle

Calculating the centroid slip angle of the k +1 th generation of the target particle

Calculating the k-th generation yaw rate actual value of the target particle

And centroid slip angle of kth generation of the target particle

Updating to the k +1 th generation actual value of the yaw rate of the target particle

And centroid slip angle of the k +1 th generation of the target particle

3. The particle swarm expandable control torque constraint method according to claim 2, characterized in that the inertial weight τ of the target particle is updated as the number of iterations changes, the calculation method of the inertial weight comprising the following steps:

obtaining a maximum weight τ of the target particle _max Minimum weight τ of the target particle _min ；

According to the iteration times t of the target particles _gen Calculating an inertial weight τ of the target particle:

4. the particle swarm extension based control torque constraint method according to claim 1, wherein the calculation method of the linear region boundary of the actual centroid slip angle gain of the automobile comprises the following steps:

obtaining the fitting parameters a required by the fitting calculation ₁₀ And fitting parameters b required for fitting calculation ₁₀ Fitting parameters c required for fitting calculation ₁₀ Fitting parameters d required for fitting calculation ₁₀ ；

According to the speed v of the automobile _x Calculating the front wheel steering angle linear zone boundary of the automobile by a fitting algorithm, namely the maximum value delta of the front wheel steering angle of the automobile _max ：

The maximum value delta of the front wheel steering angle of the automobile _max And calculating the linear region boundary of the actual mass center sideslip angle gain of the automobile through an automobile two-degree-of-freedom model.

5. The particle swarm extension based control torque constraint method according to claim 1, wherein the domain of the feature quantity is judged through a measure pattern, and the division method of the measure pattern comprises the following steps:

acquiring a correlation function K (S) of the position and the distance of the characteristic quantity in the extension control domain;

taking the measure mode of the characteristic quantity in the classical domain as M ₁ ：M ₁ ＝{S|K(S)＞1}；

Taking the measurement mode of the characteristic quantity in the extension domain as M ₂ ＝{S|0＜K(S)≤1}；

Taking the measurement mode of the characteristic quantity in the non-domain as M ₃ ＝{S|K(S)＜0}。

6. The particle swarm scalable control torque constraint method according to claim 5, wherein the design method of the correlation function comprises the following steps:

obtaining an extension area interval X of the wheel _k And a classical domain interval X of said wheel _j ；

Of the extensible control fieldConnecting and extending the original point and the characteristic quantity to obtain a connecting line, and intersecting the connecting line and the boundary of the extension domain to obtain an intersection point Q ₂ Intersection point Q ₃ Said point of intersection Q ₂ Has an abscissa of-s ₁ Said point of intersection Q ₃ Has an abscissa of s ₁ 。

Calculating the extension distance rho (Q, X) from the characteristic quantity to the extension domain _k )：

Calculating the correlation function K (S):

7. the particle swarm scalable control torque constraint method according to claim 6, wherein when the feature quantity is in a measure mode M ₁ When the maximum driving torque of the wheel is limited only by the peak torque of the in-wheel motor, regardless of the torque constraint limit of the wheel slip, i.e. the maximum driving torque of the wheel is limited only by the peak torque of the in-wheel motor

Wherein, P _max The rated power of the hub motor is shown, and n is the rotating speed of the motor.

8. The particle swarm extension based control torque constraint method of claim 6, wherein when the feature quantity is in a measure mode M ₂ The value T is adjusted according to the driving torque in consideration of the wheel slip _e Calculating a maximum drive torque constraint of the wheel as T _max ＝[K(S)+1]T _e 。

9. The particle swarm scalable control torque constraint method according to claim 6, wherein when the feature quantity is in a measure mode M ₃ When the wheel slip is severe, the vehicleThe actual mass center slip angle enters a destabilizing state, and the maximum value of the wheel driving torque is constrained to be T _max ＝T _e 。

10. Particle swarm extension based control torque restraint system for providing torque restraint for a plurality of wheels of an automobile, characterized in that it employs the particle swarm extension based control torque restraint method according to any of claims 1-9, comprising:

the extension control domain coordinate system calculation module is used for acquiring a slip rate value s of the wheel and an actual centroid slip angle beta of the automobile as characteristic quantities to form Q (s, beta), and establishing a coordinate system of an extension control domain by taking the slip rate value s as a horizontal coordinate and the actual centroid slip angle beta as a vertical coordinate;

an extension control domain region division module, configured to perform region division on the extension control domain:

classical domain: ((-s) ₁ ,s ₁ )，(-β ₁ ,β ₁ ))；

And (3) extension domain: ((-s) ¹ ₁ ,s ¹ ₁ )，(-β ¹ ₁ ,β ¹ ₁ ) In which-s) ¹ ₁ ∈(-s ₂ ,-s ₁ )，s ¹ ₁ ∈(s ₁ ,s ₂ )，-β ¹ ₁ ∈(-β ₂ ,-β ₁ )，β ¹ ₁ ∈(β ₁ ,β ₂ )；

Non-domain: ((-s) ² ₁ ,s ² ₁ )，(-β ² ₁ ,β ² ₁ ) In which-s) ² ₁ ∈(-∞,-s ₂ )，s ² ₁ ∈(s ₂ ,∞)，-β ² ₁ ∈(-∞,-β ₂ )，β ² ₁ ∈(β ₂ ,∞)；

A classical domain boundary calculation module for obtaining a linear region boundary of the actual centroid slip angle gain of the automobile as a boundary beta of a vertical coordinate of the classical domain ₁ Obtaining instituteSlip threshold s of said wheel _max As the abscissa boundary s of the classical domain ₁ I.e. s ₁ ＝s _max ；

An extension control domain boundary calculation module, configured to calculate an optimal slip ratio value s of the wheel by a particle swarm algorithm using the slip ratio value s of the current state of the wheel and the actual centroid slip angle β of the current state of the automobile ₂ And the slip rate optimum s ₂ Corresponding optimal centroid slip angle β ₂ ；

The characteristic quantity domain judging module is used for judging whether the characteristic quantity is in a classical domain or not, and if so, implementing the torque constraint of the classical domain on the wheels of the automobile; when the characteristic quantity is not in the classical domain, judging whether the characteristic quantity is in an extension domain, if so, implementing torque constraint of the extension domain on the wheels of the automobile; when the characteristic quantity is not in the extension area, the characteristic quantity is in the non-area, and the non-area torque constraint is carried out on the wheels of the automobile.

Images