CAD Method Supporting Mechanical Innovative Design Based on

Artificial Intelligence

Gao Honghui

Yantai Vocational College, Yantai, Shandong, 264670, China

Keywords: Artificial Intelligence, Mechanical Innovative Design, CAD Method, Agent Technology.

Abstract: With the development of CAD technology, modern CAD design software is not only used to replace manual

drawing, but also plays a more and more important role in enterprises. The current focus of CAD is to adapt

to the market demand. The purpose of this paper is to study the CAD method of mechanical innovative design

based on artificial intelligence. The traditional design method based on finite element is limited by the huge

gap between CAD and CAE and the huge time-consuming caused by multiple numerical calculations, so it is

difficult for engineers to modify the CAD model and carry out CAE structural analysis quickly and accurately.

Combining the innovative research of CAD design with computer, a new CAD design method based on

artificial intelligence is developed, which provides a high-performance system for researchers by using the

high storage capacity of computer. In this paper, firstly, the mechanical innovative design method is

summarized, and then through the study of the relevant design theory of transmission parts, the object-oriented

design method is adopted. Aiming at the shortcomings of CAD system, the multi-agent is deeply studied

Based on the application of CAD technology, evolutionary technology and collaborative design technology

in CAD system, a CAD system supporting innovative conceptual design is established and implemented.

Finally, through the system test and result analysis, the static test error is 0, and the two errors of the dynamic

test line are fixed. Finally, the system runs well, and the genetic algorithm has good adaptability to the layout

design.

1 INTRODUCTION

In order to enhance their competitiveness, shorten the

product development cycle and reduce the

development cost of products, enterprises constantly

explore and practice some advanced integrated

manufacturing technology and methods of electronic

computer to adapt to the increasingly fierce market

competition (Oliveira, Rbd , et al, 2019). At the same

time, after the merger and reform of global companies

or enterprises, manufacturing industry has developed

into a global industry of large-scale, wide-ranging

and wide-ranging cooperation (Kuna and Skaan,

2019). Many famous manufacturing companies and

enterprises from the world have not only R & D

and manufacturing departments, but also relevant

parts manufacturing enterprises. Therefore, many

complex products need to be co designed and

constructed by producers and designers distributed in

different locations and locations in design (Tao and

Li, et al. 2018). All enterprises are actively using

CAD, computer - aided design ).

This paper mainly relies on the academic

background of the provincial mechanical engineering

discipline, and studies the graduate innovation

research project (cx2016b079) and the national key

basic research technology development plan ("973"

plan), The research is carried out under the guidance

and support of 2010cb328005, the key project of

national natural science fund construction

(61232014), and the general project of NSFC

construction (11472101). In order to overcome the

huge gap between CAD and CAE in mechanical

structure design and the huge effort needed in the

process of repeated calculation and modification of

structure, the geometric model of structural CAD is

realized by engineers, which is simple, convenient,

quick and accurate (Chen, J. L and C. L. Lee, 2017).

Based on the professional background of the

mechanical engineering discipline, this paper

conducts relevant research work under the support of

the provincial graduate innovation and research

project (cx2016b079), the national key basic research

and development plan (973 Plan, 2010cb328005), the

244

Honghui, G.

CAD Method Supporting Mechanical Innovative Design Based on Artiﬁcial Intelligence.

DOI: 10.5220/0013539000004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 1, pages 244-249

ISBN: 978-989-758-763-4

National Natural Science Fund key project

(61232014), and the National Natural Science

Foundation general project (11472101) ( Mandal, D.

K and C. S. Syan, 2016). In order to overcome the

huge gap between CAD and CAE in mechanical

structure design and the huge time required for

repeatedly calculating and modifying the structure,

the engineer can easily and accurately represent and

modify the geometric model of structural CAD, and

obtain the dream of the results of CAE analysis of

structure quickly and even in real time, so as to

accelerate the product development and improve

social benefits (Innovative Materials, 2016).

2 CAD METHOD FOR

MECHANICAL INNOVATIVE

DESIGN

2.1 Agent Technology

Multi agent system is one of artificial intelligence,

which has the characteristics of intelligence. In this

system, each agent is an independent and intelligent

entity, which takes action or solves problems through

interaction with external environment and

coordination between agents. Multi agent system is

usually applied in dynamic, distributed and intelligent

environment (Hadhami, et al, 2020).

2.2 The IGA-IFU Method is Used for

Accurate Reanalysis of Structural

Integration

Based on IgA, the integration of CAD and CAE is

constructed, which lays the foundation for the

subsequent real-time analysis and optimization. The

engineer can eliminate the representation error by

accurately representing the geometric model of the

structure, and easily modify the CAD geometry

model of mechanical structure by changing the

control point (Dietterich T G, 2017). The CAE model

can be obtained immediately, which avoids the

complicated conversion between CAD and CAE in

traditional methods, and the errors and time

consuming (Raedt and Kersting, et al. 2016). When

the structural modification does not affect the overall

displacement of the structure, the independent

coefficient method has a very high precision; while

when the overall displacement of the structure is

greatly affected by the structural modification,

especially when the structural modification causes the

structural deformation mode to change, the accuracy

of the independent coefficient method may not be

guaranteed (Seyedmahmoudian, et al, 2016).

3 SUPPORT THE DESIGN OF

CAD METHOD SYSTEM FOR

MECHANICAL INNOVATIVE

DESIGN

3.1 Multi Agent System

3.1.1 Question Raising

Technological innovation is the spirit and soul of

CAD. The competition of new products based on

knowledge is becoming the focus of the competition

of Chinese enterprises in the complicated global

manufacturing environment in the 21st century

(Bryson, and Winfield, 2017). At present, all kinds of

manufacturing and processing equipment

manufacturing enterprises in China have entered a

period of rapid development oriented by market main

body. The economic benefits and quality of

enterprises depend on the revolutionary innovation of

technology and products to a large extent. Whether an

enterprise has the ability to research and develop a set

of innovative products that meet the needs of the

market and the consumption mentality of different

types of customers in the coming years will become

the final decision on the survival and development of

an enterprise (Price, and Flach, 2017).

3.1.2 Multi Agent System Structure

The agent in the design environment is usually called

design agent. Design agent is a kind of computer

software that helps designers to complete the design

task in some way.

A multi agent system is generally composed of the

following aspects:

1) Multiple existing agents;

2) There is a joint intention among multiple

agents, that is, multiple agents act together to achieve

common goals;

3) Common sense: that is, the common knowledge

between agents;

4) The environment on which agent depends is the

basic guarantee for agent behavior.

Multi Agent System Supporting Innovation

Concept Design

In this system, complex design is completed by

multiple agents. Each agent has its own independent

knowledge and design decision-making scheme, and

CAD Method Supporting Mechanical Innovative Design Based on Artiﬁcial Intelligence

245

can understand the design state representation, so as

to assist us human design technology experts to

achieve the design objectives to be done. The design

strategy of agent mainly depends on some basic

algorithms, such as genetic algorithm and

classification algorithm. Once a new task is arrived,

the task decomposition agent (TDA) is used to

decompose the whole design task into many

independent sub task sets, and a product design tree

is used to represent the decomposition results. TDA

knowledge base contains many product design tree

templates. TDA selects the appropriate template

according to its product category and recommends it

to designers. The designers and engineers make

preliminary decisions and send the results to the

design agent of each component. For example, if a

user needs to submit a request for housing design,

TDA will extract the information of all components

of the house and the information of all finished houses

in a corresponding knowledge base, and decompose

the functions of these products into several relatively

independent functional components, which are

transferred to different components and design agents

for design. The design agent of each component helps

designers and personnel to do their work according to

their own tasks. When all the sub tasks are completed,

all the design results are submitted to the assembly

agent. In the process of assembly, the assembly agent

needs to check the assembly limit. For components

that do not meet the requirements, the information

that must be modified is sent to the required design

agent. After the assembly agent completes the

assembly of new products, the new products are

uploaded to the customer, and the users will evaluate

the quality of the products and give the new products

a score. If the user is satisfied with the new product,

the new product will be output. Otherwise, the

components shall be replaced or changed according

to the user's requirements. After the component is

modified, it is re submitted to the assembly agent for

assembly until the user is satisfied.

3.1.3 Specific Implementation of Agents

The implementation of component design agent

adopts genetic algorithm, which uses mathematical

functions to generate two-dimensional curves and

three-dimensional entities to inspire people's

thinking, so as to realize innovative design.

3.1.4 Improvement of Genetic Algorithm

This paper first extracts the basic features, expresses

it by component tree, then cuts and combines them,

uses genetic algorithm based on natural selection and

evolution principle, simulates the learning process

naturally, adjusts the elements of product

composition structure, and makes it have a more

efficient and reasonable structure, so as to optimize or

increase the function of products and realize the

support for innovation.

3.2 Belt Drive Design Calculation

Subsystem

Design Calculation of V-Belt

1) Main failure modes and design criteria of V-

belt drive

The main failure modes of V-belt drive are

slipping, fatigue damage, wear and static breaking of

the belt; the design criterion of V-belt drive is to

ensure that the belt has sufficient fatigue strength and

life without slipping; there are two design constraints

of V-belt drive, which are as follows:

F = 1000





≤𝐹



（1−







）

)

(1) is a condition that the mechanical transmission

will not slip, where f is the friction coefficient, α is

the angle of the pulley, e is the bottom of the natural

logarithm, and V is the linear velocity of the belt. The

fatigue strength conditions are shown in (2):

𝜎



=𝜎



+𝜎



+𝜎



≤

[

𝜎

]

)

Where 𝜎



is the tight edge stress, 𝜎



is the

bending stress, 𝜎



is centrifugal stress.

(2) Parameters and conditions for interactive input

or selection in V-belt design

The parameters and conditions that must be input

or selected interactively in the design of V-belt pulley

mainly include: starting type, load property and

working hours per day of belt drive, power generated

by belt drive, slip friction ratio, requirements of other

conditions, such as bearing size, center distance limit,

etc, When designing the pulley structure, it is

necessary to know the types of various driving force

generators.

3.3 Chain Drive Design Calculation

Subsystem

Chain drive is a kind of transmission widely used in

mechanical transmission. It transmits motion and

force by the meshing between sprocket teeth and

chain link. The design content of chain drive design

and calculation module includes roller chain and

tooth chain. The following is the theoretical basis of

INCOFT 2025 - International Conference on Futuristic Technology

246

chain drive module design and calculation and the

realization of roller chain design and calculation.

Design and Calculation of Roller Chain

3.3.1 Main Failure Modes

The main failure of chain drive is chain failure. The

main failure modes are: chain compression failure,

chain collision fracture, chain wear, and chain

crushing due to overload.

3.3.2 Design of Chain Drive

Because the roller chain itself is a standard part, it

only needs to select the corresponding type of roller

chain according to the design, so the main content of

chain drive design is to calculate the relevant

parameters and the design of sprocket.

3.4 Design and Calculation Subsystem

of Gear Transmission

Gear is one of the most important transmission parts,

which is widely used in mechanical industry,

automobile, aircraft, shipbuilding and other

industries. In the gear design calculation module, the

design of involute spur gear, helical spur gear and

straight bevel gear is studied and designed.

3.5 Data Processing

In the design calculation module, because the system

involves many kinds of parts, there are many kinds of

coefficients and data that need to be queried in the

part design. How to query the parameters manually

and obtain them efficiently and quickly in the design

module through programming is a key problem to be

solved in the design process of this system.

Mathematical Model of Data Processing

3.5.1 Least Square Method

For some given data （𝑥



，𝑦



）i=0,1,…,n In

the selected function type, find f (x) belonging to the

selected function type_ i) , make𝑒



=𝑓



𝑥





−𝑦



The

sum of squares of I is the smallest

𝑒













=

𝑓



𝑥





−𝑦















(3)

The value of 𝑓



𝑥





minimum.

Program Processing Of Data Table

For the table that needs to query parameters in the

design process (hereinafter referred to as the number

table), the processing method is: the data with small

amount is directly programmed; the data with large

amount is written into the program in the form of

determined table name, field name and field type

according to Microsoft Office Access The structure

of the database is structured storage, and the tables

involved in each module are stored as a database file.

1) Single parameter table processing

For the data table with only one independent

variable and a small amount of data, the data can be

directly programmed into the array, and the required

data can be obtained by querying according to the

change of the variable_ The value of 𝑘



is shown in

Table 1.

Table 1: Row number coefficient of multi row chain

Row

number

1 2 3 4 5

𝑘



1 1.7 2.5 3.3 4.1

Figure 1: Row number coefficient of multi row chain

As shown in Figure 1, the coefficient of row

number increases with the increase of row number.

When programming, define a one-dimensional array,

store the data in the table in the array, define an

integer variable I, according to the row number

change selected by the user, according to the change

of I value, you can query the array to obtain the row

number coefficient 𝑘



2) Interpolation query

In the design process, some tables do not list all

the data. When variables can not be directly queried

in the table to obtain the required data, interpolation

query is needed to obtain the required results. As

shown in Table 2, it is the query table of envelop

angle coefficient

Table 2: Wrap angle coefficient

ang

180

degre

160

degre

140

degre

120

degre

100

degre

𝑘

∝

1.00 0.95 0.89 0.82 0.74 0.69

1 1.7 2.5 3.3 4.1

Row

number …

Row number

CAD Method Supporting Mechanical Innovative Design Based on Artiﬁcial Intelligence

247

Figure 2: Wrap angle coefficient

As shown in Figure 2, the envelop angle

coefficient increases with the increase of envelop

angle, and the increase range is linear.

4 TEST AND RESULT ANALYSIS

System Test

Unit test is to test the subsystem at rest and in

motion. The main contents of static test include:

1) Consistency of variable type and scope

declaration

2) Accuracy of algorithm and formula

3) Is the logic clear

4) Is the symbol consistent

5) Is the jump accurate

6) Is data transmission accurate and consistent

7) Consistency of font style and color

8) Form style consistency

9) Can the code be simplified and the running

efficiency be improved

Through the code walkthrough and structure

review, the static test found no errors. The dynamic

test selects a typical example to test all the steps of

design and calculation, and exposes the possible

hidden errors of the system by inputting illegal

parameters in the test process, and corrects them in

time. After the completion of the subsystem test, each

subsystem is managed and controlled through a

unified interface. The display of each subsystem is

controlled through the management database, and

each subsystem controls its own display by accessing

the database. The single subsystem program cannot

run when the main interface of the system is not

running.

5 CONCLUSIONS

In this paper, an accurate numerical method based on

variable thickness is proposed. The effectiveness,

reliability and efficiency of the method are verified by

illustration, derivation and analysis, as well as by

typical examples and actual mechanical structures in

the field of automotive engineering. Although the

research of CAD system supporting innovation is still

in the research stage, the improvement and

improvement of system integrity, flexibility,

supporting innovation and collaboration brought by it

are remarkable, so its development prospect is broad.

Although some stage research results have been

achieved, there are still many new problems to be

further explored due to the constraints of time and

energy.

REFERENCES

Oliveira, Rbd , et al. "Supporting Information Innovative

design for the enhancement of lithium lanthanum

titanate electrolytes." Crystal Growth And Design

19.9(2019):4897-4901.

Kuna P , Skaan M , A Haková. Didactic materials

supporting CAD/CAE system teaching[J]. Luis Gómez

Chova, 2019:págs. 1641-1647.

Tao J , Li L , Yu S . An innovative eco-design approach

based on integration of LCA, CADCAE and

optimization tools, and its implementation

perspectives[J]. Journal of Cleaner Production, 2018,

187(JUN.20):839-851.

Chen, J. L. , and C. L. Lee . "Developing Sustainable

Innovative Products for the Bottom of the Pyramid by

Biomimetic Design Concepts." Procedia Cirp

61(2017):629-634.

Mandal, D. K. , and C. S. Syan . "[Lecture Notes in

Mechanical Engineering] CAD/CAM, Robotics and

Factories of the Future || Design and Modelling of Dual

Faceplate Centrifugal Casting Equipment for

Manufacturing of Turbine Bearing." 10.1007/978-81-

322-2740-3.Chapter 51(2016):523-534.

Integrating Innovation in Architecture (Design, Methods

and Technology for Progressive Practice and Research)

|| Innovative Materials. 2016,

10.1002/9781119164807:20-59.

Hadhami, et al. "Proposal of new eco-manufacturing

feature interaction-based methodology in CAD phase."

The International Journal of Advanced Manufacturing

Technology 106.3(2020):1057-1068.

Dietterich T G . Steps Toward Robust Artificial

Intelligence[J]. Ai Magazine, 2017, 38(3):3.

Raedt L D , Kersting K , Natarajan S , et al. Statistical

Relational Artificial Intelligence: Logic, Probability,

and Computation[J]. Synthesis Lectures on Artificial

Intelligence and Machine Learning, 2016, 10(2):1-189.

0.95

0.89

0.82

0.74

0.69

WRAP ANGLE

COEFFICIENT

WRAP ANGLE

INCOFT 2025 - International Conference on Futuristic Technology

248

Seyedmahmoudian, M. , et al. "State of the art artificial

intelligence-based MPPT techniques for mitigating

partial shading effects on PV systems – A review."

Renewable & Sustainable Energy Reviews

64.oct.(2016):435-455.

Bryson, J. , and A. Winfield . "Standardizing Ethical

Design for Artificial Intelligence and Autonomous

Systems." Computer 50.5(2017):116-119.

Price, S. , and P. A. Flach . "Computational support for

academic peer review: a perspective from artificial

intelligence." Communications of the Acm

60.3(2017):70-79.

CAD Method Supporting Mechanical Innovative Design Based on Artiﬁcial Intelligence

249