Method and device for on-line identification of wheel cornering stiffness

Abstract

The invention provides a method for on-line identification of wheel cornering stiffness, and the method comprises the following steps: S1, building a transverse kinematics model of a vehicle about to-be-identified parameters, the to-be-identified parameters comprising front wheel cornering stiffness and rear wheel cornering stiffness of the vehicle; s2, discretizing the transverse kinematics model of the vehicle to obtain a discrete model about the to-be-identified parameters; and S3, adjusting the output of the discrete model by means of a water circulation algorithm so as to determine the to-be-identified parameter on line. The invention further provides a device for identifying the cornering stiffness of the wheel on line and a computer program product. According to the method, the heuristic algorithm is adopted to identify the wheel cornering stiffness on line, the time-varying property of the wheel cornering stiffness is considered, and the stability control of the vehicle is remarkably improved.

Description

Technical Field

[0001] The present invention relates to a method for online identification of wheel cornering stiffness, and also relates to a device and a computer program product for online identification of wheel cornering stiffness. Background Art

[0002] Wheel cornering stiffness is a key parameter that affects vehicle stability control, and its accuracy is crucial to vehicle handling performance. As a time-varying parameter affected by many factors, online identification of wheel cornering stiffness is quite challenging. Traditional offline testing methods cannot reflect the dynamic changes of parameters in real time, resulting in inaccurate identification results. Although there are some online identification methods, they still have deficiencies in accuracy and real-time performance, and cannot meet the needs of vehicle dynamic simulation and stability control. Therefore, developing a method that can accurately identify wheel cornering stiffness online is of great significance to improving vehicle safety and handling. Summary of the invention

[0003] The purpose of the present application is to provide a method for online identification of wheel cornering stiffness, a device for online identification of wheel cornering stiffness and a computer program product, so as to at least solve some of the problems in the prior art.

[0004] According to a first aspect of the present application, a method for online identification of wheel cornering stiffness is provided, the method comprising the following steps:

[0005] Step S1, establishing a lateral kinematics model of the vehicle with respect to parameters to be identified, wherein the parameters to be identified include the front wheel cornering stiffness and the rear wheel cornering stiffness of the vehicle;

[0006] Step S2, discretizing the lateral kinematics model of the vehicle to obtain a discrete model of the parameters to be identified; and

[0007] Step S3, adjusting the output of the discrete model by means of a water cycle algorithm to determine the parameters to be identified online.

[0008] The present application particularly includes the following technical concepts: The present solution adopts a heuristic algorithm to solve the model output, which effectively enhances the algorithm's search capability and improves its adaptability to complex problems. Through the online identification scheme, the time-varying nature of the wheel cornering stiffness is taken into account, and the accuracy of parameter identification is also ensured to be unaffected by the size of the initial value. At the same time, since the modeling error can be corrected online, a more accurate mathematical model is provided for the lateral dynamic response of the vehicle. In general, the method of the present application significantly improves the stability control of the vehicle.

[0009] In an exemplary embodiment, step S1 includes: using a two-degree-of-freedom vehicle dynamics model to establish the following lateral kinematics model: M·ay =F yf +F yr , where M is the mass of the vehicle, a y is the lateral acceleration of the vehicle, F yf is the lateral force of the front wheel of the vehicle, F yr is the rear wheel lateral force of the vehicle; the lateral kinematics model is parameterized based on the wheel linear lateral force model to obtain the following lateral kinematics model about the parameters to be identified: a y =C αf b f +C αr b r , where a y is the lateral acceleration of the vehicle, b f is the first coefficient, b r is the second coefficient, C αf is the front wheel cornering stiffness, C αr is the rear wheel cornering stiffness.

[0010] In an exemplary embodiment, parameterizing the lateral kinematics model based on the wheel linear lateral force model includes: establishing a wheel linear lateral force model based on the vehicle's wheel cornering stiffness and wheel slip angle; establishing a wheel slip angle model based on the front wheel turning angle and the wheel speed angle; establishing a wheel speed angle model based on the vehicle wheelbase, the vehicle yaw rate, the vehicle lateral speed and the vehicle longitudinal speed; integrating the wheel linear lateral force model, the wheel slip angle model and the wheel speed angle model into the lateral kinematics model to express the vehicle's lateral acceleration as a function of the vehicle's front wheel cornering stiffness and rear wheel cornering stiffness.

[0011] In an exemplary embodiment, step S3 includes: inputting the error value between the actual dynamic response data and the predicted dynamic response data of the vehicle into the water circulation algorithm to iteratively adjust the estimated value of the parameter to be identified with the help of the water circulation algorithm, wherein the predicted dynamic response data of the vehicle is the output of the discrete model based on the estimated value of the parameter to be identified; and, after each iteration, checking whether a preset iteration stop condition is met, if the preset iteration stop condition is not met, continuing the iteration, and controlling the estimation of the parameter to be identified in the next iteration cycle according to the error value between the actual dynamic response data and the predicted dynamic response data of the vehicle in the current iteration cycle; when the iteration stop condition is met, stopping the iteration and outputting the estimated value of the parameter to be identified in the current iteration cycle as the optimal solution.

[0012] In an exemplary embodiment, by constructing the following fitness function C j To measure whether the preset iteration stop condition is reached:

[0013]

[0014] Wherein, e(k) is the error value between the estimated value of lateral acceleration and the actual value of lateral acceleration at the kth sampling moment, is the estimated value of lateral acceleration, a y (k) is the actual value of lateral acceleration.

[0015] In an exemplary embodiment, in step S3, the following iterative process is performed with the help of a water cycle algorithm: step S31, initializing population parameters and generating initial population individuals, each individual in the population represents a potential solution to a set of parameters to be identified; step S32, calculating the fitness value of each individual in the population and classifying all individuals in the population into streams, rivers and oceans according to the fitness value; step S33, iteratively optimizing the population by updating the positions of streams, rivers and oceans; and, step S36, when a preset iteration stop condition is reached, outputting the parameters to be identified represented by the individual with the highest fitness value in the final population as the optimal solution.

[0016] In an exemplary embodiment, in step S3, the iterative process also includes: step S34, when the evaporation conditions are met, triggering the evaporation process to reduce the number of streams and / or rivers, performing a rainfall process to randomly generate new individuals to join the population; and, step S35, adaptively updating the evaporation conditions as the iterative process proceeds.

[0017] In an exemplary embodiment, step S3 also includes: dynamically adjusting the internal parameters of the water circulation algorithm according to the vehicle driving state, driving environment conditions, calculation delay requirements and / or calculation accuracy requirements, and the internal parameters include at least one of the following: the number of initial population individuals, the maximum number of iterations, the fitness function and / or the evaporation conditions.

[0018] According to a second aspect of the present application, a device for online identification of wheel cornering stiffness is provided, the device comprising a processor and a memory, the memory storing computer program instructions, and when the computer program instructions are executed by the processor, the processor is capable of executing the method described in the first aspect of the present application.

[0019] According to a third aspect of the present application, a computer program product having a program code unit is provided, wherein the program code unit is configured to cause the computer to execute the method according to the first aspect of the present application when the computer program product is run on a computer or stored on a computer-readable storage medium. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present application will be described in more detail below with reference to the accompanying drawings, so that the principles, features and advantages of the present application can be better understood. The accompanying drawings include:

[0021] Figure 1 A flow chart of a method for online identification of wheel cornering stiffness according to an exemplary embodiment of the present application is shown;

[0022] Figure 2 A schematic diagram of a two-degree-of-freedom vehicle dynamics model according to an exemplary embodiment of the present application is shown;

[0023] Figure 3 A principle flow chart of a water circulation algorithm according to an exemplary embodiment of the present application is shown;

[0024] Figure 4 A logic block diagram of online identification of wheel cornering stiffness using a water circulation algorithm is shown; and

[0025] Figure 5 A block diagram of a device for online identification of wheel cornering stiffness according to an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0026] In order to make the technical problems, technical solutions and beneficial technical effects to be solved by the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and multiple exemplary embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the scope of protection of the present application.

[0027] Figure 1 A flow chart of a method for online identification of wheel cornering stiffness according to an exemplary embodiment of the present application is shown. The method comprises step S1, step S2 and step S3.

[0028] In step S1 , a lateral kinematics model of a vehicle with respect to parameters to be identified is established, wherein the parameters to be identified include the cornering stiffness of the front wheels and the cornering stiffness of the rear wheels of the vehicle.

[0029] The cornering stiffness of a wheel is defined as the ratio between the lateral force of the wheel and its corresponding slip angle. It is a key parameter to measure the handling stability of the wheel. The magnitude of the cornering stiffness directly affects the handling performance and stability of the vehicle. Specifically, a wheel with a higher cornering stiffness can provide a vehicle with better handling stability. This means that when turning, the wheel can resist the lateral force more effectively. Therefore, the cornering stiffness is an important consideration in vehicle dynamics analysis and wheel design.

[0030] In order to establish a lateral kinematic model, relevant driving parameters and physical parameters can be obtained from the vehicle's sensors, for example. Driving parameters include but are not limited to the vehicle's lateral speed, longitudinal speed, yaw rate and front wheel angle, which can be obtained through devices such as lateral speedometers, longitudinal speedometers, yaw ratemeters and steering angle sensors. Physical parameters include, for example, the vehicle's wheelbase (i.e., the distance from the vehicle's center of mass to the front and rear axles) and vehicle mass, which can be directly obtained by consulting the vehicle specification manual, or measured and estimated in real time through a vehicle dynamic measurement system (such as wheel speed sensors, vehicle mass distribution sensors, etc.).

[0031] Here, the exemplary lateral kinematics model establishment process can refer to the following combined Figure 2 For the sake of brevity, the description is not repeated here.

[0032] In step S2, the lateral kinematics model of the vehicle is discretized to obtain a discrete model of the parameters to be identified.

[0033] This process involves converting the continuous-time model into a discrete-time model. Commonly used discretization methods include: forward Euler method, backward Euler method, zero-order hold method, etc.

[0034] In one embodiment, for example, the continuous lateral dynamics model can be converted into a discrete time model using the zero-order hold method as a discrete model. This method assumes that the input remains constant in each sampling period, and converts the model into a discrete time model by using the model state at the current moment in each discrete time step to approximately represent all states in the entire step. Let k be the current moment, then the discretized vehicle lateral kinematics model (discrete model) can be expressed as:

[0035] a y (k) = C αf (k) b f (k)+C αr (k) b r (k)

[0036] Among them, b f (k) and b r (k) is the first coefficient and the second coefficient based on the current moment k, which integrate the driving parameters and physical parameters of the vehicle, a y (k) represents the lateral acceleration of the vehicle at the current time k, and C αf (k) and C αr (k) represent the front wheel cornering stiffness and rear wheel cornering stiffness at time k, respectively.

[0037] Through discretization, complex continuous systems can be transformed into models suitable for digital control and real-time calculation, which is of great significance for the subsequent online parameter identification process with the help of water circulation algorithm. The discrete model is not only easy to implement, but also can provide real-time prediction and feedback of the vehicle's lateral dynamic behavior, thereby significantly improving the accuracy and efficiency of online identification of wheel cornering stiffness.

[0038] Step S3, adjusting the output of the discrete model by means of a water cycle algorithm to determine the parameters to be identified online.

[0039] The water cycle algorithm is a bio-inspired algorithm designed to solve multi-objective optimization problems. It mimics the dynamic process of water cycle in nature, especially how streams and rivers eventually flow into the ocean. In this process, water enters the atmosphere through evaporation, forms clouds, and condenses and falls under appropriate conditions, completing the water cycle. The water cycle algorithm draws on this natural phenomenon and applies the transfer mechanism of water between the surface, ocean and atmosphere to parameter optimization. It explores the solution space and finds the optimal solution by simulating the flow and conversion process of water. This method not only has a powerful global search capability, but also can adaptively adjust the search strategy to adapt to complex optimization problems.

[0040] Unlike one-time static identification, the core of online identification lies in its continuity and dynamism. With the help of the water cycle algorithm, the output of the discrete model can be adjusted iteratively and combined with the actual vehicle feedback data to continuously optimize the solution results of the parameters to be identified. This dynamic parameter identification process can not only capture the time-varying characteristics of the wheel cornering stiffness, but also provide more accurate handling performance during vehicle driving, thereby realizing real-time monitoring and adjustment of vehicle dynamic behavior.

[0041] The principle flow and logic implementation of the water cycle algorithm will be combined below Figure 3 and Figure 4 Further explanation is omitted for brevity. Through this method, key parameters such as vehicle cornering stiffness can be accurately identified, providing a scientific basis for optimizing vehicle handling performance.

[0042] Figure 2 A schematic diagram of a two-degree-of-freedom vehicle dynamics model according to an exemplary embodiment of the present application is shown.

[0043] The modeling process of the two-degree-of-freedom vehicle dynamics model involves describing the vehicle's motion in two degrees of freedom: lateral (y-axis) and yaw (rotation around the z-axis). Figure 2 As shown, first establish a coordinate system with the center of mass of the vehicle as the origin. In this coordinate system, the x-axis points to the longitudinal direction of the vehicle, and the y-axis points to the lateral direction of the vehicle. In lateral motion, the lateral force F acting on the front wheel of the vehicle is mainly considered. yfand the lateral force F acting on the rear wheel of the vehicle yr According to Newton's second law, for lateral motion, the following lateral kinematic model of the vehicle can be established:

[0044] M.a y =F yf +F yr

[0045] Where M is the mass of the vehicle, a y is the lateral acceleration of the vehicle, F yf is the lateral force of the front wheel of the vehicle, F yr is the lateral force on the rear wheels of the vehicle.

[0046] Next, the vehicle lateral kinematic model can be parameterized (simplified) using the wheel linear lateral force model to obtain the vehicle's lateral acceleration a y About the parameter C to be identified αf and C αr The function expression is as follows:

[0047] Assuming that the lateral force of the wheel is linearly related to the wheel slip angle, the linear lateral force model of the wheel can be established as follows:

[0048] The front wheel lateral force model of the vehicle is:

[0049] F yf =2C αf α f

[0050] Among them, F yf is the lateral force on the front wheels of the vehicle, C αf is the front wheel cornering stiffness, α f is the front wheel slip angle.

[0051] The rear wheel lateral force model of the vehicle is:

[0052] F yr =2C αr α r

[0053] Among them, F yr is the lateral force on the rear wheels of the vehicle, C αr is the rear wheel cornering stiffness, α r is the rear wheel slip angle.

[0054] In addition, it can be assumed that the wheel slip angle is directly related to the front wheel steering angle and the wheel speed angle, so the wheel slip angle model can be established based on the front wheel steering angle and the wheel speed angle as follows:

[0055] The front wheel slip angle model of the vehicle is:

[0056] αf =δ-θ vf

[0057] Among them, α f is the front wheel slip angle of the vehicle, δ is the front wheel turning angle of the vehicle, and θ vf is the front wheel speed angle.

[0058] The rear wheel slip angle model of the vehicle is:

[0059] α r =-θ vr

[0060] Among them, α r is the rear wheel slip angle of the vehicle, θ vr is the rear wheel velocity angle of the vehicle.

[0061] In addition, the wheel speed angle model can be established based on the vehicle wheelbase, vehicle yaw rate, vehicle lateral velocity and vehicle longitudinal velocity as follows:

[0062] The front wheel speed angle model of the vehicle is:

[0063]

[0064] Among them, θ vf is the front wheel speed angle of the vehicle, is the lateral velocity of the vehicle, l f is the distance from the center of mass of the vehicle to the front axle (front wheelbase), ψ is the yaw rate of the vehicle, v x is the longitudinal velocity of the vehicle.

[0065] The rear wheel velocity angle model of the vehicle is:

[0066]

[0067] Among them, θ vf is the front wheel speed angle of the vehicle, is the lateral velocity of the vehicle, l r is the distance from the center of mass of the vehicle to the rear axle (rear wheelbase), ψ is the yaw rate of the vehicle, v x is the longitudinal velocity of the vehicle.

[0068] Substitute the wheel linear lateral force model, wheel velocity angle model, and wheel slip angle model into the lateral kinematics model to express the lateral acceleration of the vehicle as a function of the front wheel cornering stiffness and rear wheel cornering stiffness of the vehicle. That is, by combining the above equations, the following lateral kinematics model for the parameters to be identified can be obtained:

[0069]

[0070] This equation describes the lateral acceleration of the vehicle, ay The parameters to be identified, namely the cornering stiffness of the front and rear wheels C αf and C αr The relationship between.

[0071] Next, the first coefficient b can be extracted f and the second coefficient b r , to simplify the lateral kinematic model to:

[0072] a y =C αf b f +C αr b r

[0073] in, These coefficients b f and b r By integrating the measurable driving parameters and known physical parameters of the vehicle, the lateral acceleration a y and the parameter to be identified C αf and C αr The relationship between them is simpler and easier to handle.

[0074] Figure 3 A principle flow chart of a water circulation algorithm according to an exemplary embodiment of the present application is shown.

[0075] The following combination Figure 3 Describe how to perform an iterative process with the help of a water cycle algorithm to achieve online identification of parameters.

[0076] In step S31, the population parameters are initialized and initial population individuals are generated. Each individual in the population represents a potential solution to a set of parameters to be identified, and the total number of population individuals represents the number of potential solutions.

[0077] Specifically, population parameters may include: the total number of individuals in the initial population N pop , the number of parameters to be identified D var , the maximum number of iterations of the algorithm i max And the evaporation condition parameter d max In addition, the upper bound UB and the lower bound LB of the parameter to be identified can also be defined during the initialization process.

[0078] The size of the initial population N pop For example, depending on computing power, the more individuals in the initial population, the more scenarios are covered, and the easier it is to get the optimal solution, but the dedicated computing power for each iteration step is higher. Therefore, based on the current processor performance and simulation data, a population size value should be selected that can ensure that the algorithm effectively explores the solution space without excessively consuming computing resources.

[0079] The upper bound UB and lower bound LB of the parameters to be identified can be set according to the physical characteristics of the tire. Tire manufacturers usually provide corresponding boundary parameters according to different speed levels. These parameters can be set by appropriately enlarging the recommended values ​​in the manual, which helps to narrow the search range. In addition, the historical parameter identification results can also be used as initial values ​​to set the upper and lower bounds to improve the efficiency and accuracy of the algorithm.

[0080] The maximum number of iterations is set mainly to prevent the algorithm from falling into an infinite loop when it cannot find the optimal solution. This is an empirical parameter to ensure that the algorithm can terminate in a reasonable time.

[0081] In the initial stage, individuals in the population are not classified, and all individuals are regarded as "raindrops". The rainfall process can classify individuals into streams, rivers or oceans. var The problem of identifying parameters is that the initial population Total population can be expressed as an N pop ×D var The matrix:

[0082]

[0083] In step S32, the fitness value C of each individual in the population is calculated. j , and according to the fitness value C j Classify all individuals in the population into streams, rivers, and oceans. Such classification and initialization help the algorithm search the solution space more efficiently in subsequent iterations.

[0084] In the water cycle algorithm, the fitness value C j It is an indicator used to measure the quality or merits of a set of potential solutions. Specifically, the fitness value can be calculated according to the following formula:

[0085]

[0086] in, represents the fitness function, which is, for example, the error value between the estimated lateral acceleration and the actual vehicle dynamic response data. According to the value of the fitness function (i.e., the fitness value), all population individuals can be sorted and classified according to the fitness size to obtain oceans, rivers, and streams.

[0087] In one embodiment, it can be stipulated that the individual with the smallest fitness value is the ocean, and the number is 1; the total number of individuals in the river and the ocean is N sr ; The remaining individuals are divided into streams, the number of which is N pop -N sr . For example, the initial population Totalpopulation can be re-expressed as:

[0088]

[0089] Among them, Total population is the total population, N pop is the total number of individuals in the initial population, D var is the number of parameters to be identified.

[0090] In addition, in this step, the number of streams flowing to the ocean and the number of streams flowing to the river in the current population can be further calculated based on the fitness value. This process draws on the water convergence mechanism in nature to simulate the distribution of potential solutions in the search space.

[0091] Specifically, the following calculation can be used to determine the number of streams that flow to each river or ocean:

[0092]

[0093] Among them, NS n represents the number of streams flowing into the nth river or ocean, N streams is the total number of streams, round means rounding, N sr The Cost in the formula is the sum of the total number of individuals in rivers and oceans. n is the fitness value of the nth river or ocean, is the sum of the fitness values ​​of all rivers and oceans. This method distributes streams according to the fitness value of each river or ocean, so that rivers or oceans with higher fitness values ​​can attract more streams. In the water cycle algorithm, this allocation mechanism simulates the natural flow process of water from high to low in nature, which helps guide the algorithm to effectively search for the optimal solution.

[0094] In step S33, the population is iteratively optimized by updating the positions of streams, rivers and oceans. These updates can follow a specific mathematical formula.

[0095] Specifically, the updated position of a stream flowing into the ocean can be expressed as:

[0096]

[0097] The updated positions of streams flowing into a river can be expressed as:

[0098]

[0099] The updated position of a river flowing into the ocean can be expressed as:

[0100]

[0101] Where i is the number of iterations; rand is a uniformly distributed random number between 0 and 1 to indicate the randomness of the position update; C is a constant, usually 2, which is used to control the step size of the update. Respectively represent the current position of streams, rivers and oceans, Represent the updated positions of streams and oceans, respectively.

[0102] After updating the stream position according to the above formula, the corresponding fitness value will be calculated. If the value is better than the fitness value of the river connected to it, the position of the stream will be exchanged with the position of the river, otherwise it will not be exchanged. If a stream has a better fitness value than the river connected to it after updating its position, it means that the new position of the stream may represent a better potential solution. Similarly, there is a similar exchange mechanism between streams and oceans, and between rivers and oceans.

[0103] In this way, the algorithm can simulate the flow and circulation of water in nature, continuously and iteratively update the positions of streams, rivers, and oceans, thereby achieving population optimization. Through position exchange, the algorithm not only retains excellent solutions, but also encourages the diversity of the population and avoids premature convergence to the local optimal solution. This mechanism helps the algorithm strike a balance between global search and local search.

[0104] In step S34, when the evaporation conditions are met, the evaporation process is triggered to reduce the number of streams and / or rivers, and the rainfall process is performed to randomly generate new individuals to join the population.

[0105] In one embodiment, the evaporation condition can be set as follows: when the distance between a certain number of rivers and the ocean is less than a preset first distance threshold d max , or optionally when the distance between a certain proportion of individuals in the population and the ocean is greater than a preset second distance threshold d min This design strategy prevents the algorithm from converging prematurely near the global optimal solution and also avoids the potential solutions from being too dispersed in the solution space, thus ensuring that the algorithm can effectively search globally while performing in-depth solutions in promising areas.

[0106] In other embodiments, the evaporation condition can also be set according to the fitness value of the individual. For example, if the fitness value of the individual has not been significantly improved in several consecutive iterations, or the overall concentration of individuals in the population is too high, indicating that the algorithm may be trapped in a local optimum, the evaporation process can be triggered. In this way, the algorithm can periodically "refresh" the population and introduce new potential solutions to maintain the diversity and dynamics of the search.

[0107] After the evaporation conditions are met, new individuals can be generated through rainfall operations according to the following expression:

[0108]

[0109] in, represents newly generated raindrops (which can be further classified as streams or rivers based on their fitness values), LB is the lower bound of the parameter to be identified, UB is the upper bound of the parameter to be identified, and rand is a random number uniformly distributed between 0 and 1. Through the rainfall process, new streams and rivers can be randomly generated at different locations in the solution space. These new individuals represent new potential solutions, which help to increase the diversity of solutions and explore new search spaces.

[0110] By introducing evaporation and rainfall processes, the algorithm can avoid converging to the local optimal solution too early. In the water cycle, streams and rivers that flow too slowly and cannot reach the ocean will eventually evaporate. The emergence of the evaporation process will attract new precipitation and form new streams or rivers. This dynamic adjustment mechanism enables the algorithm to adaptively adjust the search strategy, enhancing the algorithm's global search capability and adaptability.

[0111] In step S35, the evaporation condition is adaptively updated as the iteration proceeds. For example, as the iteration proceeds, the distance threshold d may be dynamically updated according to the following formula: max :

[0112]

[0113] in, is the distance threshold of the i-th iteration cycle, is the distance threshold of the updated i+1th iteration cycle, i max is the maximum number of iterations.

[0114] As the algorithm progresses, the conditions for triggering evaporation may become more stringent. In the early stages of the iteration, larger d max The value of d allows a wider search and helps explore different areas of the solution space. As the iteration progresses, the gradually decreasing d max The value prompts the algorithm to focus on areas where there is a greater chance of an optimal solution and conduct a more detailed search. This strategy not only promotes global search, but also enhances the local search depth, increasing the possibility of finding a global optimal solution. By dynamically adjusting the distance threshold, the water cycle algorithm can flexibly switch between different search scales to meet the needs of complex optimization problems.

[0115] In the optional step S35', the internal parameters of the water circulation algorithm can also be dynamically adjusted according to the vehicle driving state, driving environment conditions, calculation delay requirements and / or calculation accuracy requirements. These internal parameters at least include the number of individuals in the initial population, the maximum number of iterations, the fitness function and / or the evaporation condition.

[0116] In a fixed driving scenario, the above internal parameters of the algorithm are relatively stable and do not need to be adjusted frequently. However, once the driving scenario changes, these parameters must be re-identified to adapt to the new driving conditions. This is the importance of online identification.

[0117] In one embodiment, when the vehicle speed changes significantly or the road conditions (such as humidity and flatness) change, the above internal parameters need to be re-evaluated and adjusted. For example, when these changes occur, the maximum number of iterations i can be appropriately increased. max and the total number of individuals in the initial population N pop To ensure that the algorithm can fully explore the solution space and find the optimal solution. In addition, if the vehicle's driving state or driving environment changes, or if the error is large and difficult to converge during the calculation process, the initial population parameters can be set in advance based on experience or experimental data (such as experimental results obtained in similar scenarios), and then based on these newly set initial population parameters, the parameters to be identified can be recalculated more effectively to adapt to the changed driving environment.

[0118] In one embodiment, when driving at high speed or in complex road conditions, more iterations are required to find more optimal wheel cornering stiffness parameters. On the contrary, in scenarios where computing resources are limited or real-time requirements are high, the maximum number of iterations can be reduced to speed up the convergence of the algorithm.

[0119] In one embodiment, for applications that require fast response, it may be necessary to adjust the total number of individuals in the initial population of the algorithm or reduce the number of iterations to reduce the calculation time.

[0120] Through this dynamic adjustment mechanism, the water cycle algorithm can respond to different driving scenarios more flexibly and achieve optimal performance under various conditions. This adaptive parameter adjustment strategy not only improves the robustness of the algorithm, but also enhances its practicality in practical applications.

[0121] In step S36, when the preset iteration stop condition is reached, the parameter to be identified represented by the individual with the highest fitness value in the final population is output as the optimal solution. This will be discussed in detail below. Figure 4 The examples are further described.

[0122] Figure 4 The logical block diagram of online identification of wheel cornering stiffness using a water circulation algorithm is shown.

[0123] As above combined Figure 1 As has been introduced, the following discrete model can be obtained by discretizing the established vehicle lateral dynamics model:

[0124]

[0125] refer to Figure 4 , and They represent the front wheel cornering stiffness and rear wheel cornering stiffness estimated by the water circulation algorithm at time k, which are related to the known coefficients b of the discrete model. f (k) and b r (k) and the estimated value of the vehicle's lateral acceleration is obtained by summing That is, the output of the discrete model. In the first iteration of the algorithm, the wheel cornering stiffness obtained by identifying the vehicle when it leaves the factory can be used as the initial value of the parameter to be identified, and then the estimated value of the lateral acceleration is calculated based on this.

[0126] Based on the measurement of the actual dynamic response of the vehicle, the actual value of the vehicle's lateral acceleration a can be obtained. y (k), this actual value is used as feedback to compare with the lateral acceleration estimated by the water circulation algorithm The error value e(k) is calculated by comparing the two parameters. This error value is provided to the water cycle algorithm for the next iteration of the algorithm. The water cycle algorithm uses this error value to adjust the estimated value of the parameter to be identified. and The goal is to reduce the difference between the estimated and actual lateral acceleration values, thereby accelerating the search and convergence process for the global optimal solution.

[0127] The above iterative process will be repeated until the error value e(k) converges to a preset threshold (e.g., close to zero), which indicates that the estimated lateral acceleration is very close to the actual lateral acceleration and satisfies the iteration stop condition. At this time, the parameter to be identified represented by the individual with the highest fitness value in the final population output by the water cycle algorithm is taken as the optimal solution.

[0128] In a specific embodiment, the fitness function C can be constructed as follows j To measure whether the preset iteration stop condition is reached:

[0129]

[0130] Among them, e(k) is the error value at the kth sampling moment, that is, the difference between the estimated value and the actual value of the lateral acceleration, is the estimated value of lateral acceleration, a y (k) is the actual value of the lateral acceleration.

[0131] In other embodiments, the fitness function may also have other expressions. In addition, the iteration termination condition may also be diversified. For example, the iteration termination condition may also be set based on the amount of change of the fitness function, reaching a predetermined number of iterations, or taking into account one or more of these factors.

[0132] Through this online identification process, the water circulation algorithm is able to update the estimate of the wheel cornering stiffness in real time, thereby improving the accuracy and adaptability of vehicle stability control. This approach allows the wheel cornering stiffness to be dynamically identified during vehicle driving to cope with changing driving conditions and vehicle loads, ensuring vehicle handling performance and safety.

[0133] Figure 5 A block diagram of a device for online identification of wheel cornering stiffness according to an exemplary embodiment of the present application is shown.

[0134] like Figure 5 As shown, the device 10 includes a processor 11 and a memory 12, wherein the memory 12 stores computer program instructions. When the computer program instructions are executed by the processor, the processor 11 can, for example, execute a method for online identification of wheel cornering stiffness (which has been described above with reference to Figure 2 The computer program instructions may be stored in a computer-readable storage medium. The computer-readable storage medium may include, for example, a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card, a secure digital card, a flash memory card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. The processor may be a central processing unit, or may be other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0135] In one embodiment, the device 10 can be deployed in the cloud, and cloud infrastructure can be used to provide cloud services or calculation results to users. Users can purchase parameter identification services from cloud service providers and upload sample data or vehicle operation data using an application program interface. The device 10 deployed in the cloud will perform online identification of wheel cornering stiffness based on these data, and provide real-time feedback of online identification results to vehicle users.

[0136] In another embodiment, the device 10 may also be arranged in a vehicle to achieve localized parameter identification.

[0137] Although the specific embodiments of the present application are described in detail herein, they are provided for the purpose of explanation only and should not be considered to limit the scope of the present application. Various substitutions, changes and modifications may be conceived without departing from the spirit and scope of the present application.

Claims

1. A method for online identification of wheel cornering stiffness, the method comprising the following steps: Step S1, establishing a lateral kinematics model of the vehicle with respect to parameters to be identified, wherein the parameters to be identified include the cornering stiffness of the front wheels and the cornering stiffness of the rear wheels of the vehicle; Step S2, discretizing the lateral kinematics model of the vehicle to obtain a discrete model of the parameters to be identified; as well as Step S3, adjusting the output of the discrete model by means of a water cycle algorithm to determine the parameters to be identified online.

2. The method according to claim 1, wherein: The step S1 comprises: The following lateral kinematics model of the vehicle is established using the two-degree-of-freedom vehicle dynamics model: M·a y =F yf +F yr Where M is the mass of the vehicle, a y is the lateral acceleration of the vehicle, F yf is the lateral force of the front wheel of the vehicle, F yr is the lateral force of the rear wheel of the vehicle; The lateral kinematics model is parameterized based on the wheel linear lateral force model to obtain the following lateral kinematics model for the parameters to be identified: a y =C αf ·b f +C αr ·b r Among them, a y is the lateral acceleration of the vehicle, b f is the first coefficient, b r is the second coefficient, C αf is the front wheel cornering stiffness, C αr is the rear wheel cornering stiffness.

3. The method according to claim 2, wherein: Parameterization of the lateral kinematics model based on the wheel linear lateral force model includes: A wheel linear lateral force model is established based on the vehicle's wheel cornering stiffness and wheel slip angle; A wheel slip angle model is established based on the front wheel steering angle and wheel speed angle; A wheel speed angle model is established based on the vehicle wheelbase, the vehicle yaw rate, the vehicle lateral speed and the vehicle longitudinal speed; The wheel linear lateral force model, the wheel slip angle model and the wheel speed angle model are integrated into the lateral kinematic model to express the lateral acceleration of the vehicle as a function of the front wheel cornering stiffness and the rear wheel cornering stiffness of the vehicle.

4. The method according to any one of claims 1 to 3, wherein: The step S3 comprises: Inputting the error value between the actual dynamic response data of the vehicle and the predicted dynamic response data into the water cycle algorithm to iteratively adjust the estimated value of the parameter to be identified by means of the water cycle algorithm, wherein the predicted dynamic response data of the vehicle is the output of the discrete model based on the estimated value of the parameter to be identified; After each iteration, check whether the preset iteration stop condition is reached. If not, continue the iteration, and control the estimation of the parameters to be identified in the next iteration cycle according to the error value between the actual dynamic response data and the predicted dynamic response data of the vehicle in the current iteration cycle; when the iteration stop condition is reached, stop the iteration and output the estimated value of the parameter to be identified in the current iteration cycle as the optimal solution.

5. The method according to claim 4, wherein: By constructing the following fitness function C j To measure whether the preset iteration stop condition is reached: Wherein, e(k) is the error value between the estimated value of lateral acceleration and the actual value of lateral acceleration at the kth sampling moment, is the estimated value of lateral acceleration, a y (k) is the actual value of lateral acceleration.

6. The method according to any one of claims 1 to 5, wherein: In step S3, the following iterative process is performed with the help of the water cycle algorithm: Step S31, initializing population parameters and generating initial population individuals, each individual in the population represents a potential solution for a set of parameters to be identified; Step S32, calculating the fitness value of each individual in the population and classifying all individuals in the population into streams, rivers and oceans according to the fitness value; Step S33, iteratively optimizing the population by updating the positions of streams, rivers and oceans; as well as Step S36, when the preset iteration stop condition is reached, the parameter to be identified represented by the individual with the highest fitness value in the final population is output as the optimal solution.

7. The method according to claim 6, wherein: In step S3, the iterative process further includes: Step S34, when the evaporation condition is met, triggering the evaporation process to reduce the number of streams and / or rivers, and performing the rainfall process to randomly generate new individuals to join the population; and Step S35 , adaptively updating the evaporation conditions as the iterative process proceeds.

8. The method according to any one of claims 1 to 7, wherein: The step S3 further comprises: The internal parameters of the water circulation algorithm are dynamically adjusted according to the vehicle driving state, driving environment conditions, calculation delay requirements and / or calculation accuracy requirements, and the internal parameters include at least one of the following: the number of initial population individuals, the maximum number of iterations, the fitness function and / or the evaporation condition.

9. A device (10) for online identification of wheel cornering stiffness, the device (10) comprising a processor (11) and a memory (12), the memory (12) storing computer program instructions, and when the computer program instructions are executed by the processor (11), the processor (11) is capable of executing the method according to any one of claims 1 to 8. 10 . A computer program product having program code means, which are configured to cause a computer to execute the method according to claim 1 , when the computer program product is run on a computer or is stored on a computer-readable storage medium.

Images