Automobile spoiler optimization method for reducing wind resistance

Abstract

The invention relates to the technical field of commercial vehicles, in particular to an automobile spoiler optimization method for reducing wind resistance, and the method comprises the steps: carrying out the parameterization optimization of an automobile spoiler structure through employing a DOE experiment method, and determining the influence degree of different inputs on a target variable; specifically, when a DOE experimental method is adopted for design, the construction of a radial basis proxy function model is considered, after CS21 is used as a basis function of an RBF, a root-mean-square error method is used for screening data, and effective data are distinguished and extracted; and finally, carrying out iterative optimization on the screened data through a GA (Genetic Algorithm) to obtain an optimal solution wind resistance value, and optimizing the angle and length of the automobile spoiler according to the wind resistance value to achieve the purpose of reducing the wind resistance of the automobile.

Description

Technical Field

[0001] The invention relates to the technical field of commercial vehicles, and in particular to a method for optimizing a car rear wing for reducing wind resistance. Background Art

[0002] When commercial vehicles are driving, as the speed increases, the proportion of air resistance to the total resistance of the vehicle increases, and the largest increase is in wind resistance; therefore, for vehicles that drive on high-speed roads more often, optimizing the vehicle's wind resistance can more significantly improve the vehicle's fuel economy.

[0003] At present, some studies have shown that by optimizing the body design, reducing the drag coefficient, and adopting aerodynamic kits, the air resistance of light trucks during driving can be reduced, thereby improving fuel economy. However, in the current actual operation of body optimization design, the optimization of vehicle drag reduction accessories often requires multiple rounds of optimization simulation comparison results, and is very dependent on the optimization experience of engineers, which greatly increases R&D time and costs. At the same time, the existing market has too single optimization methods for its vehicle accessories, and cannot accurately and efficiently optimize the air resistance of the entire vehicle to improve fuel economy performance. Summary of the invention

[0004] The purpose of the present invention is to provide a method for optimizing a car rear wing for reducing wind resistance, which aims to achieve further reasonable optimization of the car rear wing structure by parameterizing the modeled car rear wing structure, designing and building a proxy model through a DOE experimental method, and then performing optimization calculations through a GA genetic algorithm, so as to finally achieve the purpose of improving the fuel economy of commercial vehicles.

[0005] To achieve the above object, the present invention provides a method for optimizing a car rear wing for reducing wind resistance, comprising the following steps:

[0006] Step 1: Design the tail wing structural parameters based on the DOE experimental method;

[0007] Step 2: Construct a proxy model based on radial basis function;

[0008] Step 3: Optimizing the agent model based on GA genetic algorithm;

[0009] Step 4: Perform global optimization on the optimization iteration results and derive the tail wing structure parameters;

[0010] Step 5: Verify and calculate the tail wing structure parameters.

[0011] Optionally, in step 1, the Hammersley sampling method is used, and the sampling steps are as follows:

[0012] Step 1.1: Construct the Hammersley sequence point and use different digit combinations to represent a non-negative integer p. The expression based on base R is as follows:

[0013] p=p m p m-1 …p2p1p0

[0014] p=p0+p1R+p2R 2 +…+p m R m

[0015] Where m = [log R P], P i =[0,R-1], [ ] is the rounding function;

[0016] Step 1.2: Based on base R -1 , reverse the order of each digit of the integer p to obtain the interval [0,1] n The N Hammersley sequence points in are expressed as follows.

[0017]

[0018] Among them, φ(P) refers to the Van der Corput sequence, which is used to generate uniformly distributed points;

[0019] Step 1.3: The constructed experimental sample points are the n-dimensional design space [0,1] n The N Hammersley sequence points in are expressed as follows:

[0020]

[0021] Where, p=0,1,2,...,N-1; R1,R2,R3,...,R n-1 The values ​​are n-1 prime numbers in sequence.

[0022] Optionally, step 2 selects a basis function of the radial basis function, and the construction process is as follows:

[0023] Step 2.1: Use CS21 function as the basis function, the expression is:

[0024]

[0025] Step 2.2: Use the root mean square error method to evaluate the accuracy of the established proxy model.

[0026] Optionally, the optimization steps based on the GA genetic algorithm are as follows:

[0027] Step 3.1: Select the constructed radial basis function as the proxy model;

[0028] Step 3.2: Set the number of iterations of the proxy model;

[0029] Step 3.3: Set the number of candidate solutions to be evaluated at each iteration to determine the population size required for iteration;

[0030] Step 3.4: Set the mutation rate of the corresponding population;

[0031] Step 3.5: Set the elite strategy and distribution index of the population;

[0032] Step 3.6: Set the global search level and search process preference.

[0033] Optionally, the execution process of step 4 is specifically to find the optimal wind resistance value from the iterative data group of step 3, and after combining the upper and lower limits of component parameter optimization, derive the vehicle rear wing structure parameters to obtain the front and rear angles and lengths before and after optimization.

[0034] The invention provides an automobile tail wing optimization method for reducing wind resistance. The method adopts a DOE experimental method to perform parameterized optimization of an automobile tail wing structure, and determines the influence of different input quantities on a target variable. Specifically, when the DOE experimental method is used for design, a proxy function is constructed, and after using CS21 as a basis function of the proxy function, a root mean square error method is used to screen data, so as to achieve identification and extraction of effective data. Finally, the screened data is iteratively optimized through a GA genetic algorithm to obtain an optimal solution wind resistance value, and the angle and length of the automobile tail wing are optimized according to the wind resistance value, so as to achieve the purpose of reducing the automobile wind resistance. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 The present invention is a schematic flow chart of the steps of a method for optimizing a car rear wing for reducing wind resistance.

[0037] Figure 2 Schematic diagram of the residual error of the proxy model according to a specific embodiment of the present invention.

[0038] Figure 3 It is a schematic diagram of genetic algorithm iteration according to a specific embodiment of the present invention.

[0039] Figure 4 It is a schematic diagram of the angle and length of the tail wing structure before and after optimization of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0040] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0041] The abbreviations of the English terms appearing in this application have the following meanings:

[0042] DOE: Design of Experiments, experimental design;

[0043] GA: Genetic Algorithm, an optimization search algorithm based on natural selection and genetics principles;

[0044] RBF: Radial Basis Function, radial basis function proxy model;

[0045] CFD: Computational Fluid Dynamics.

[0046] See also Figure 1 The present invention provides a method for optimizing a car rear wing for reducing wind resistance, comprising the following steps:

[0047] S1: Design tail wing structural parameters based on DOE experimental method;

[0048] S2: Constructing a proxy model based on radial basis functions;

[0049] S3: Optimizing the agent model based on GA genetic algorithm;

[0050] S4: Perform global optimization on the optimization iteration results and derive the tail wing structure parameters;

[0051] S5: performing verification calculation on the tail wing structure parameters.

[0052] The following is further described in conjunction with specific embodiments and execution steps:

[0053] In step S1, first, based on the DOE experimental design method, mathematical statistics and orthogonal principle are combined to reasonably design the experimental space, and the influence of different input variables on the target variable is determined through experiments. After analysis and comparison, the present invention adopts the Hammersley sampling method, and its sampling steps are as follows:

[0054] 1. Construct Hammersley sequence points and use different digit combinations to represent a non-negative integer p. Its expression based on base R is as follows:

[0055] p=p m p m-1 …p2p1p0

[0056] p=p0+p1R+p2R 2 +…+p m R m

[0057] Where m = [log R P], P i =[0,R-1], [ ] is the rounding function.

[0058] 2.Based on base R -1 , reverse the order of each digit of the integer p to obtain the interval [0,1] n The N Hammersley sequence points in are expressed as follows.

[0059]

[0060] 3. The constructed experimental sample points are the n-dimensional design space [0,1] n The N Hammersley sequence points in are expressed as follows:

[0061]

[0062] Where, p=0,1,2,...,N-1; R1,R2,R3,...,R n-1 The values ​​are n-1 prime numbers in sequence.

[0063] Step S2: constructing a proxy model based on radial basis function, the construction process is as follows:

[0064] 1. Using CS21 function as the basis function of radial basis function, its expression is:

[0065]

[0066] 2. Evaluate the accuracy of the established model to ensure the predictive ability of the surrogate model using the root mean square error method (RMSE);

[0067] The root mean square error method is defined as:

[0068]

[0069] When using this method, it is necessary to: (1) filter the data and remove outliers;

[0070] (2) Try different model types, such as linear regression, decision tree, neural network, etc., and choose the model that best suits the current problem;

[0071] (3) Perform certain smoothing corrections on the model prediction results;

[0072] (4) Take multiple measurements and calculate the average value.

[0073] 3. Observe the results. In the RMSE formula, a set of n input values ​​given to the model is denoted as y i , is the predicted value of the surrogate model, is the mean of the true response. When the calculated RMSE result approaches 0, its accuracy is higher. Figure 2 The residual graph of the proxy model established for the present invention has residuals within 0.2, which has high accuracy.

[0074] In step S3, if Figure 3 As shown, the optimization is performed based on the GA genetic algorithm, and the optimization steps are as follows:

[0075] 1. Select a suitable proxy model. In the present invention, RBF is selected as the proxy model to replace the simulation with high computational cost.

[0076] 2. Set the number of iterations of the proxy model to 250, which means that the proxy model will be iterated 250 times.

[0077] 3. Set the size of the population to be iterated. Here, the number of candidate solutions evaluated in each iteration is set to 114.

[0078] 4. The mutation rate of the above population is set, and the probability of individual gene mutation is 0.01, which means that the probability of data mutation in the above data set is one percent.

[0079] 5. Set the elite strategy of the above population to 10% of the population and the distribution index to 5 to influence the step length or range of the variation.

[0080] 7. Set the global search level to high, and set it to tend to explore the solution space during the search process to find the global optimal solution.

[0081] Step S4, after 100 iterations of the genetic algorithm, the optimal wind resistance value is found from the existing 29053 sets of data, and its value is 0.41524. As shown in Table 1, after retaining only four decimal places, the optimal solution is the 28748th set of data. After combining the upper and lower limits of the component parameter optimization, the car rear wing is derived, and its angle and length before and after optimization are shown in Figure 4 As shown, Figure 4 The left side shows the original angle and length, and the right side shows the optimized angle and length.

[0082] Table 1 Genetic algorithm optimization results

[0083]

[0084] In step S5, a CFD simulation calculation is performed on the wind resistance system of the whole vehicle, and its wind resistance value is 0.417, which is 22% higher than that of the original vehicle and is 0.4% different from the result of the proxy prediction model, meeting the expected target.

[0085] In summary, compared with the prior art, the advantages of the present invention are:

[0086] 1. The parametric optimization method designed with the DOE test method can more effectively and accurately optimize the parameters of the commercial vehicle tail wing.

[0087] 2. Most of the existing technologies do not consider optimizing the rear wing of commercial vehicles by constructing a proxy model. By using the method of constructing a proxy model, the experimental cost can be reduced while ensuring accuracy.

[0088] 3. Most of the existing technologies do not consider the use of genetic algorithms to automatically optimize the basis functions of the proxy model. The present invention uses the GA genetic algorithm for optimization, finds the optimal solution wind resistance value in the data obtained by iterative calculation, and optimizes the angle and length of the car's rear wing according to the wind resistance value to achieve the reduction of the car's wind resistance.

[0089] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A method for optimizing a car rear wing to reduce wind resistance, characterized in that: The following steps are involved: Step 1: Design the tail wing structural parameters based on the DOE experimental method; Step 2: Construct a proxy model based on radial basis function; Step 3: Optimizing the agent model based on GA genetic algorithm; Step 4: Perform global optimization on the optimization iteration results and derive the tail wing structure parameters; Step 5: Verify and calculate the tail wing structure parameters.

2. The method for optimizing a car rear wing to reduce wind resistance according to claim 1, characterized in that: In step 1, the Hammersley sampling method is used, and the sampling steps are as follows: Step 1.1: Construct the Hammersley sequence point and use different digit combinations to represent a non-negative integer p. The expression based on base R is as follows: p=p m p m-1 ...p2p1p0 p=p0+p1R+p2R 2 +...+p m R m Where m = [log R P], P i =[0, R-1], [ ] is the rounding function; Step 1.2: Based on base R -1 , reverse the order of each digit of the integer p to obtain the interval [0, 1] n The N Hammersley sequence points in are expressed as follows: Among them, φ(P) refers to the Van der Corput sequence, which is used to generate uniformly distributed points; Step 1.3: The constructed experimental sample points are the n-dimensional design space [0, 1] n The N Hammersley sequence points in are expressed as follows: Where x n (p) is the Radical Inverse function based on base b, which reverses the digits of the base b representation of the integer p and then interprets it as a decimal between 0 and 1, p = 0, 1, 2, ..., N-1; R1, R2, R3, ..., R n-1 The values ​​are n-1 prime numbers in sequence.

3. The method for optimizing a car rear wing to reduce wind resistance according to claim 2, characterized in that: Step 2 constructs the radial basis function. The construction process is as follows: Step 2.1: Use CS21 function as the basis function, the expression is: Step 2.2: Use the root mean square error method to evaluate the accuracy of the established proxy model.

4. The method for optimizing a car rear wing to reduce wind resistance according to claim 3, characterized in that: The optimization steps based on GA genetic algorithm are as follows: Step 3.1: Select the constructed radial basis function as the proxy model; Step 3.2: Set the number of iterations of the proxy model; Step 3.3: Set the number of candidate solutions to be evaluated at each iteration to determine the population size required for iteration; Step 3.4: Set the mutation rate of the corresponding population; Step 3.5: Set the elite strategy and distribution index of the population; Step 3.6: Set the global search level and search process preference.

5. The method for optimizing a car rear wing to reduce wind resistance according to claim 4, characterized in that: The execution process of step 4 is specifically to find the optimal wind resistance value from the iterative data group of step 3, and after combining the upper and lower limits of component parameter optimization, derive the structural parameters of the car rear wing to obtain the front and rear angles and lengths before and after optimization.

Images