A vehicle material data dimension reduction method and system for deep learning technology

Abstract

This invention discloses a method and system for dimensionality reduction of vehicle material data for deep learning technology. The method includes the following steps: Step 1, obtaining a list of vehicle material grades; Step 2, encoding the vehicle material grades; Step 3, defining a one-to-one correspondence between material grade codes and material grades; Step 4, opening the vehicle model; Step 5, obtaining a list of all vehicle parts; Step 6, updating the vehicle color according to the material grade codes. This invention represents materials by color, reducing the dimensionality of vehicle material parameters from hundreds of thousands to hundreds of levels, achieving effective dimensionality reduction of vehicle material data.

Description

Technical Field

[0001] This invention belongs to the field of AI model construction technology for vehicle performance development, specifically relating to a method and system for dimensionality reduction of vehicle material data for deep learning technology. Background Technology

[0002] Deep learning is widely used in image recognition, discovering distributed feature representations of data by learning the inherent patterns and hierarchical representations of sample data. In vehicle performance development, deep learning technology can be used to build AI models that learn the mapping relationship between vehicle data and vehicle performance, enabling rapid prediction of vehicle performance and intelligent optimization of vehicle data. Given the same algorithm and computing power, data is the key factor determining the performance of an AI model. A dataset, as a core component of data resources, refers to data specifically designed, collected, cleaned, labeled, and managed for training artificial intelligence algorithm models.

[0003] Currently, the main drawback of deep learning applications in vehicle performance development is the insufficient sample data. The primary means of addressing this issue is dimensionality reduction; dimensionality reduction can reduce model complexity, especially in situations with insufficient sample data, ultimately improving the performance of the trained AI model. Vehicle data includes structural and material data. Regarding material data, there are hundreds of types of vehicle materials, each with varying performance parameters; statistically, a certain passenger vehicle may have as many as 100,000 performance parameters for all its materials. Given the insufficient vehicle sample data, effective dimensionality reduction of the material data is necessary. Two main dimensionality reduction methods have been published in the literature:

[0004] ①Method 1: The AI ​​model uses vehicle image data as input. All images are grayscale images, and material differences are not considered at all.

[0005] ②Method 2: The AI ​​model uses vehicle image data as input. The image uses three colors to represent material types (red for plastic, green for glass, and blue for metal), without considering the differences after subdividing the three types of materials.

[0006] The two methods mentioned above can effectively reduce the dimensionality of vehicle material data, reducing vehicle material parameters from the tens of thousands level to the single level; however, they cannot accurately distinguish vehicle materials, which seriously affects the performance of AI models. Summary of the Invention

[0007] To address the above issues, this invention proposes a method and system for dimensionality reduction of vehicle material data using deep learning technology. By representing materials with colors, the dimensionality of vehicle material parameters is reduced from the tens of thousands level to the hundreds level, thus achieving effective dimensionality reduction of vehicle material data.

[0008] The objective of this invention is achieved through the following technical solution:

[0009] As a first aspect of the present invention, a method for dimensionality reduction of vehicle material data for deep learning technology is provided, comprising the following steps:

[0010] Step 1: Obtain the vehicle material license plate number list;

[0011] Step 2: Encode the vehicle material grade;

[0012] Step 3: Define the one-to-one correspondence between material grade codes and material grades;

[0013] Step 4: Open the vehicle model;

[0014] Step 5: Obtain a complete parts list for the vehicle;

[0015] Step 6: Update the vehicle color according to the material grade code.

[0016] Furthermore, step two includes:

[0017] 2.1) The vehicle material grade code includes three digits: units, tens, and hundreds.

[0018] 2.2) Each digit has a value range of [0-9], with ten possible values; there are a total of 1000 possible combinations.

[0019] Furthermore, step three includes:

[0020] 3.1) Construct a JavaScript object using the JavaScript language, denoted as colors;

[0021] 3.2) The key for colors is the material grade, and the value is the material grade code;

[0022] 3.3) Object property access method: object[key].

[0023] Furthermore, step four involves opening the vehicle model in a simulation design software, which can be any one of PRIMER, Hypermesh, ANSA, or CATIA.

[0024] Furthermore, step five includes:

[0025] 5.1) Use "var m_p = Model.Select('Select model')" to select the vehicle model and assign the value to the variable m_p;

[0026] 5.2) Use “var ps=Part.GetAll(m_p)” to get all the parts of the vehicle model m_p and assign them to the variable ps.

[0027] Furthermore, step six includes:

[0028] 6.1) Use "ps.length > 0" to get and check the length of ps. If the length of ps is greater than 0, proceed to the next step.

[0029] 6.2) Take the first part from ps and denote it as p;

[0030] 6.3) Calculate the material grade of p;

[0031] 6.4) Calculate the material grade code of p, and denote the hundreds, tens, and digits of the code as x, y, and z, respectively;

[0032] 6.5) Calculate the R, G, and B components of the color corresponding to the material grade code;

[0033] 6.6) Update the color of part p.

[0034] Furthermore, the method used in step 6.2) is as follows: use "var p = ps[0]" to assign the first part of ps to the variable p; use "ps.shift()" to delete the first part of ps.

[0035] Furthermore, the method used in step 6.3) is to obtain the material grade of part p using "p.mid".

[0036] Further, step 6.4) employs the following method: Calculate the material grade code colors[p.mid] corresponding to the material grade p.mid of part p using "var rgb = colors[p.mid]", and assign it to the variable rgb; assign the hundreds digit of the code to the variable x using "var x = rgb.substring(0,1)", and assign the hundreds digit of the code to the variable x using "var y = ...

[0037] "rgb.substring(1,2)" assigns the tens digit of the encoding to the variable x, and "var z=rgb.substring(2,3)" assigns the units digit of the encoding to the variable x.

[0038] As another aspect of the present invention, a vehicle material data dimensionality reduction system for deep learning technology is also provided, comprising:

[0039] The vehicle material grade list acquisition module is used to acquire the vehicle material grade list;

[0040] The vehicle material grade coding module is used to obtain the vehicle material grade coding.

[0041] The material grade and material grade code correspondence module is used to establish a one-to-one correspondence between vehicle material grades and vehicle material grade codes;

[0042] The vehicle model retrieval module is used to open vehicle models in simulation design software;

[0043] The vehicle parts list acquisition module is used to obtain a complete parts list for a vehicle model.

[0044] The parts color update module is used to update the vehicle color according to the material grade code.

[0045] The present invention has the following beneficial effects:

[0046] ①This invention can reduce vehicle material parameters from the level of tens of thousands to the level of hundreds.

[0047] ②This invention can accurately distinguish the types of vehicle materials.

[0048] ③This invention can automate the color representation of vehicle material types.

[0049] ④ This invention can ensure the uniqueness of vehicle material grade codes.

[0050] ⑤ This invention can automatically calculate the color based on the material grade code and can guarantee the uniqueness of the color. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the content of the embodiments of the present invention and these drawings without creative effort.

[0052] Figure 1 This is a flowchart of the vehicle material data dimensionality reduction method for deep learning technology described in an embodiment of the present invention;

[0053] Figure 2 This is a diagram illustrating the vehicle model described in an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram illustrating the principle of the material color calculation method described in an embodiment of the present invention. Detailed Implementation

[0055] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0056] Example 1

[0057] like Figure 1 As shown, a method for dimensionality reduction of vehicle material data using deep learning technology, taking a passenger car as an example, includes the following steps:

[0058] Step 1: Obtain the vehicle material grade list. The vehicle material grade list is shown in Table 1:

[0059] Table 1 List of Vehicle Material Grades

[0060]

[0061]

[0062] Step 2: Encode the vehicle material grades. The list of vehicle material grade codes is shown in Table 2:

[0063] Table 2 List of Vehicle Material Grade Codes

[0064]

[0065]

[0066] The vehicle material grade coding method is as follows:

[0067] 2.3) The encoding includes three digits: units, tens, and hundreds.

[0068] 2.4) The value range of each digit is [0-9], with ten possible values; there are a total of 1000 combinations, which can represent 1000 kinds of materials, meeting the coding requirements of commonly used vehicle materials (more than 100 kinds).

[0069] Step 3: Define the one-to-one correspondence between material grade codes and material grades. The steps are as follows:

[0070] 3.4) Construct a JavaScript object using the JavaScript language, denoted as colors.

[0071] 3.5) The JavaScript object `colors` is a container for key-value pairs, which are usually written as `name:value` (with the key and value separated by a colon). In this invention, the key of `colors` is the material grade, and the value is the material grade code. For example, `'16':'000'` means that the key is the material grade '16' and the value is the material grade code '000'.

[0072] 3.6) The method for accessing object properties is: object[key]; for example, colors['16'] can access the material grade code '000' corresponding to material grade '16'.

[0073] Step 4: Open the vehicle model:

[0074] Open the PRIMER software. In PRIMER, select and open the vehicle model via "Model|Read". The vehicle model is as follows: Figure 2 As shown.

[0075] Step 5: Obtain a complete parts list for the vehicle:

[0076] 5.3) Use “var m_p=Model.Select('Select model')” to select the vehicle model and assign the value to the variable m_p.

[0077] 5.4) Use “var ps=Part.GetAll(m_p)” to get all parts of the vehicle model m_p and assign them to the variable ps.

[0078] Step Six: Update the vehicle color according to the material grade code:

[0079] 6.7) Use "ps.length>0" to get and judge the length of ps. If the length of ps is greater than 0, proceed to the next step; otherwise, end.

[0080] 6.8) Take the first part from ps and denote it as p. The method is as follows: use "var p = ps[0]" to assign the first part of ps to the variable p; use "ps.shift()" to delete the first part of ps.

[0081] 6.9) Calculate the material grade of p. The method is to use "p.mid" to obtain the material grade of part p.

[0082] 6.10) Calculate the material grade code of p, and denote the hundreds, tens, and digits of the code as x, y, and z, respectively.

[0083] The method is as follows: Use "var rgb = colors[p.mid]" to calculate the material grade code colors[p.mid] corresponding to the material grade p of part p, and assign it to the variable rgb; use "var x = rgb.substring(0,1)" to assign the hundreds digit of the code to the variable x, use "var y = rgb.substring(1,2)" to assign the tens digit of the code to the variable x, and use "var z = rgb.substring(2,3)" to assign the units digit of the code to the variable x.

[0084] 6.11) Calculate the R, G, and B components of the color corresponding to the material grade code.

[0085] The method is as follows: R = x * 25, G = y * 25, B = z * 25; since the values ​​of x, y, and z range from 0 to 9, the values ​​of R, G, and B range from 0 to 225, which meets the channel intensity value range requirements of the RGB color mode; the principle of the material color calculation method is as follows. Figure 3 As shown, the material-color reference list is shown in Table 3:

[0086] Table 3 Material-Color Reference List

[0087]

[0088]

[0089] 6.12) Update the color of part p;

[0090] The method is to use "p.colour=Colour.RGB(r,g,b)".

[0091] 6.7) Go back to 6.1) Repeat the above steps to perform a batch update of vehicle colors.

[0092] Example 2

[0093] The difference between this embodiment and Embodiment 1 is that step four, opening the vehicle model, is implemented in other simulation / design software, such as Hypermesh, ANSA, CATIA, etc.

[0094] Example 3

[0095] A system for dimensionality reduction of vehicle material data using deep learning technology, comprising:

[0096] The vehicle material grade list acquisition module is used to acquire the vehicle material grade list;

[0097] The vehicle material grade coding module is used to obtain the vehicle material grade coding.

[0098] The material grade and material grade code correspondence module is used to establish a one-to-one correspondence between vehicle material grades and vehicle material grade codes;

[0099] The vehicle model retrieval module is used to open vehicle models in simulation design software;

[0100] The vehicle parts list acquisition module is used to obtain a complete parts list for a vehicle model.

[0101] The parts color update module is used to update the vehicle color according to the material grade code.

[0102] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for dimensionality reduction of vehicle material data for deep learning technology, characterized in that, Includes the following steps: Step 1: Obtain the vehicle material license plate number list; Step 2: Encode the vehicle material grade; Step 3: Define the one-to-one correspondence between material grade codes and material grades; Step 4: Open the vehicle model; Step 5: Obtain a complete list of vehicle parts; Step Six: Update the vehicle color according to the material grade code; Step Six includes: use Get and determine the length of ps. If the length of ps is greater than 0, proceed to the next step. Take the first part from ps and denote it as p; Calculate the material grade of p; Calculate the material grade code of p, and denote the hundreds, tens, and digits of the code as x, y, and z, respectively; Calculate the R, G, and B components of the color corresponding to the material grade code; Update the color of part p.

2. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, Step two includes: The vehicle material grade code consists of three digits: the ones digit, the tens digit, and the hundreds digit. The value range for each digit is: There are ten possible values; there are a total of 1000 possible combinations.

3. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, Step three includes: Use the JavaScript language to construct JavaScript objects, denoted as colors; The key for colors is the material grade, and the value is the material grade code; Object property access method: object[key].

4. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, The fourth step is to open the vehicle model in a simulation design software, which can be any one of PRIMER, hypermesh, ANSA, or CATIA.

5. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, Step five includes: use Select a vehicle model and assign it to the variable m_p; use Obtain vehicle model All parts are identified and assigned to the variable ps.

6. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, The method used in step 6.2) is as follows: using Assign the first part from ps to the variable p; use Delete the first part in Photoshop.

7. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, The method used in step 6.3) is as follows: using Obtain the material grade of part p.

8. The method for dimensionality reduction of vehicle material data for deep learning technology as described in claim 1, characterized in that, The method used in step 6.4) is as follows: using Calculate the material grade code corresponding to the material grade p.mid of part p. And assign the value to the variable rgb; use Assign the hundreds digit of the code to the variable x, and use... Assign the tens digit of the code to the variable x, and use... Assign the units digit of the code to the variable x.

9. A vehicle material data dimensionality reduction system for deep learning technology, used to implement the vehicle material data dimensionality reduction method as described in any one of claims 1-8, characterized in that, The vehicle material data dimensionality reduction system includes: The vehicle material grade list acquisition module is used to acquire the vehicle material grade list; The vehicle material grade coding module is used to obtain the vehicle material grade coding. The material grade and material grade code correspondence module is used to establish a one-to-one correspondence between vehicle material grades and vehicle material grade codes; The vehicle model retrieval module is used to open vehicle models in simulation design software; The vehicle parts list acquisition module is used to obtain a complete parts list for a vehicle model. The parts color update module is used to update the vehicle color according to the material grade code.

Images