Sie sind auf Seite 1von 15

The Optimization of Two-Stage Planetary Gear Train

Based on Mathmatica

Tianpei Chen, Zhengyan Zhang, Dingfang Chen, and Yongzhi Li

Wuhan University of Technology, Hubei Wuhan 430063


jasongervin@163.com, zzy0309@yahoo.com.cn,
dfchen@whut.edu.cn, whlgd.lyz@263.net

Abstract. Planetary gear reducer has a lot of advantages,such as high


transmission and efficiency, compact structure, and has a variety of applications
in construction machinery and equipment, hoisting and conveying machinery
and so on,The optimization design of the planetary gear train could make the
volume at minimum(as well as the weight at minimum) under the conditions of
carrying capacityThis paper focus on the optimization of two stage planetary
gear train with the differential evolution algorithm, based on Mathmatica. The
author established mathematical model and source program is present in this
paper. After the optimization, the author verifies the optimal result, including
the contact fatigue stress and tooth bending strength fatigue stress. The
verification infers that the optimization based on Mathmatica with differential
evolution algorithm is effective and correct.

Keywords: Mathmactica, optimization and design, planetary gear reducer,


verification and validation of gear.

1 Introduction

Planetary gear trains take a very important place among the gear transmissions which
are used in many branches of industry. Under similar operating conditions, planetary
gear train have a number of advantages as comparing to the standard transmission
with shafts: light weight, compact size, large speed ratio, high efficiency, long service
life, low noise and so on. Therefore a special attention should be taken to its
design [1].
The optimal design of planetary gear trains is a complex engineering problem and
a multi-objective optimization problem because of very complicated calculations, a
great number of conditions, geometrical and functional constraints and a large number
of solution variants. In this paper, based on Mathmatica, we used differential
evolution algorithm in the optimization of two stage planetary gear train.

Q. Zu, B. Hu, and A. Eli (Eds.): ICPCA-SWS 2012, LNCS 7719, pp. 122136, 2013.
Springer-Verlag Berlin Heidelberg 2013
The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 123

2 The Optimal Model of Planetray Gear Train

For the given kinematics scheme of the planetary gear train (Figure1), kinematics
parameters of the input shaft and the overall gear ratio, find a solution (basic
constructional parameters as number of teeth, modules), which simultaneously the
following objective functions: Weight of the planetary gear train.

Fig. 1. Planetary gear train

The vector of design variables is identified as follows:

Z = { x1 , x2 , x3 , x4 , x5 ; x6 , x7 , x8 , x9 } = {Z1 , m1 , b1 , X 1 , i1 ; Z1' , m2 , b2 , X 2 } (1)


T T

Where: Z1 , Z1' are the numbers of teeth of the first stage sun gear and the second
stage sun gear , m1 , m2 are the modules of the gear wheels respectively, b1 , b2 are
the tooth width of the gear respectively, X 1 , X 2 are the modification coefficient of
the sun gear respectively, i is the transmission ratio of the first stage.

2.1 Determine the Objective Function

To design the planetary gear reducer with minimal weight, the total volume of all sun
gears and planetary gears can be taken as the objective function.


F= b1 ( d a21 + 3d a22 ) + b2 ( d a' 12 + 3d a' 2 2 ) (2)
4
124 T. Chen et al.

Where: d a1 , d a 2 , d a1' , d a 2' are the addendum circle diameter of the gear wheels
respectively, b1 , b2 are the tooth width of the gear respectively.
Simultaneously, these variables are involved in the following equations.

d a = ( Z + 2ha ) m (3)

So the objective function is

1
2

f ( x) = x3 ( ( x1 + 2(1 + x4 ) ) x2 ) + 3 ( x5 2) x1 + 2(1 + x4 ) x2
2

4 2
(4)
1 u
2

+ x8 ( ( x6 + 2(1 + x9 ) ) x7 ) + ( 2) x6 + 2(1 + x9 ) x7
2

2 x5

Where: u is the overall transmission ratio of the planetary gear train.

2.2 Determine Constrain Conditions

The constraint conditions are expressed as follows.

(a) Contact fatigue stress conditions:

2 K1T1 (u1 + 1)
1 = Z H 1 Z E1 Z 1 HP1
b1d12u1
(5)
2 K 2T2 (u2 + 1)
1 = Z H 2 Z E 2 Z 2
' HP 2
b2 d12' u1

Where: Z H --- node region coefficient, Z E ---elastic coefficient, Z ---contact ratio


coefficient; d1 , d1' are the pitch diameter of the sun gear
respectively; HP1 , HP 2 are the permitted contact fatigue stress of the sun gear
respectively; b1 , b2 are the tooth width of the gear respectively; u1 , u2 are the ratio of
tooth number respectively; T1 , T2 are the input torque respectively; K1 , K 2 are the load
factor respectively and load factor K = K A KV K K , where K A --- application
The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 125

factor; KV ---dynamic factor, K --- load distribution factor, and K --- valve of
load partition factor.

(b) Tooth bending strength fatigue stress conditions:

2 K1T1
F1 = YFa1YSa1Y 1 FP1
b1d1m1
(6)
2 K 2T2
F1
' = YFa 2YSa 2Y 2 FP 2
b2 d1' m2

Where: YFa --- tooth form factor, YSa ---stress concentration coefficient, Y ---contact
ratio coefficient; b1 , b2 are the tooth width of gear respectively. d1 , d1' , are the
pitch diameter of the sun gear respectively; m1 , m2 are the modules of the gear
wheels respectively ; T1 , T2 are the input torque respectively; FP1 , FP 2 are the
permitted bending fatigue stress of the sun gear respectively; K1 , K 2 are the load
factor respectively.

(c) Transmission ratio condition

According to Mechanical Principle, the number of annular gear and sun gear must
meet the following equations:

Z3
i1 = +1
Z1
(7)
Z 3'
i2 = +1
Z1'

(d) Meshing of the gears

According to Mechanical Principle, the sum of tooth number of annular gear and sun
gear has to be integral multiple of the number of planetary gear.

Z1 + Z 3
= int
3
(8)
Z1' + Z 3'
= int
3
126 T. Chen et al.

(e) Parallelism of distance between axes of the input and output shafts

Z1 + Z 2 Z 3 Z 2
=
cos 12' cos 23
'

(9)
Z1' + Z 2' Z 3' Z 2'
=
cos 1''2' cos 2' '3'

Where: cos 12' , cos 23


'
, cos 12'' ' , cos 2' '3' are the meshing angles of the gears
respectively.

(f) Adjacent conditioned

d a 2 < 2a12' sin60


(10)
d a 2' < 2a1''2' sin60
' '
Where: d a 2 , d a 2' are the outside diameters of the planetary gear; a12 , a12' ' are the
actual center distance of gear wheels 1 and 2 , gear wheels 1' and 2' .
(g) Module conditions

2 m1 9;
2 m2 9 (11)

(h) Tooth width conditions

b1
0.6 Z 1.3Z
m1
(12)
b
0.6 Z 2 1.3Z
m2

3 Optimal Example

Given the total transmission ratio is 55.46, the total input torque of the first
stage T1 = 5.69 105 N mm and the total input torque of the second
stage T2 = 4.78 106 N mm , gear material is 20CrMnMo and heat treatment is
quenching surface and its rigidity is HRC57-61.
According to Mechanical Design Handbook, the coefficients above are selected
respectively.
The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 127

3.1 The First Stage

Node region coefficient Z H 1 = 2.22 , elastic coefficient Z E1 = 189.98 MPa ,


contact ratio coefficient Z 1 = 0.95 , K1 = 2.89 , tooth form factor YFa1 = 2.29 ,
stress correction coefficient YSa1 = 1.73 , contact ratio coefficient Y 1 = 1.12 , the
permitted contact fatigue stress HP1 = 1033.41MPa , the permitted bending fatigue
stress FP1 = 499.39 MPa .

3.2 The Second Stage

Node region coefficient Z H 1 = 2.25 , elastic coefficient Z E1 = 189.98 MPa ,


contact ratio coefficient Z 1 = 0.94, K1 = 2.95 , tooth form factor YFa1 = 2.32 ,
stress correction coefficient YSa1 = 1.73 , contact ratio coefficient Y 1 = 1.08 , the
permitted contact fatigue stress HP1 = 1104.47 MPa , the permitted bending fatigue
stress FP 2 = 521.53MPa .

4 The Realization of Optimization Based on Mathmatica and


Results

4.1 Simplification

In order to solve the question better, with high speed and high efficiency, we
simplified the constraint conditions (the modules, the tooth number, and the tooth
width) in smaller range artificially. The advantage is less iterations, higher solving
speed, and higher efficiency in the solution of the mathematical model.

z1 13
2m 9
1

55 b1 61

4 x 1 (13)
17 1


5 i1 50
z1' 13

2 m2 9

110 b2 140
4
x2 1
17
128 T. Chen et al.

4.2 The Method of Differential Evolution


The differential Evolution (DE) method of Storn and Price (1995) is perhaps the
fastest evolutionary computational procedure yielding most accurate solutions to
continuous global optimization problems. It consists of three basic steps: (i)
generation of (large enough) population with individuals in the m-dimensional
space, randomly distributed over the entire domain of the function in question and
evaluation of the individuals of the so generated population by finding f(x), where x is
the decision variable; (ii) replacement of this current population by a better fit new
population, and (iii) repetition of this replacement until satisfactory results are
obtained or the given criteria of termination are met.
The strength of DE lays on replacement of the current population by a new
population that is better fit. Here the meaning of better is in the Pareto improvement
sense. A set Sa is better than another set Sb iff : (i) no xi
Sa is inferior to the

corresponding member of xi Sb ; and (ii) at least one member xk Sa is better

than the corresponding member xk Sb. Thus, every new population is an
improvement over the earlier one. To accomplish this, the DE method generates a
candidate individual to replace each current individual in the population. A crossover
of the current individual and three other randomly selected individuals obtains the
candidate individual from the current population. The crossover itself is probabilistic
in nature. Further, if the candidate individual is better fit than the current individual, it
takes the place of the current individual else the current individual passes into the next
iteration.

Fig. 2. The method of differential evolution


The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 129

4.3 Source Program


After the treatment of objective function and constraints conditions, we input the
mathematical models to Mathmatica 7.0 and the source program is as follows:

NMinimize[
{3.1415926/4*(x3*(((x1+2*(1+x4))*x2)^2+3*((0.5*(x5-
2)*x1+2*(1+x4))*x2)^2)+x8*(((x6+2*(1+x9))*x7)^2+
((0.5*(55.46/x6-2)*x6+2*(1+x9))*x7)^2)),
(5.69*10^5)/((x3*(x1*x2)^2)*(x5/(x5-2))-1.15410,
0,
(5.69*10^5/x8*(x6*x7) ^2 *(55.46/(55.46/x5-2))-
1.29750,
0,
5.69*10^5/(x1*x2^2*x3)-19.44200,
0,
(5.69*10^5*x5)/(x6*x7^2*x8)-20.40210,
0,
- 3 /2*x1*x5+(0.5*(x5-2)*x1+2*(1+x4))*x2<0,
- 3 /2*55.46*x7/x6+(0.5*(55.46/x5-2)*x6+2*(1+x9))*x7<0,
0.6*x1*x2-x30,0,
x3-1.3*x1*x20,0,
0.6*x7*x8-x90,0,
x9-1.3*x7*x80,0,
x113,
x22,
x29,
x355,
x361,
X44/17,
x41,
5-x50,
x550,
x613,
x72,
x79,
x8110,
x8140,
X94/17,
x91},
{x1,x2,x3,x4, x5,x6,x7,x8 ,x9},
Method {"DifferentialEvolution",

"RandomSeed" 20},
MaxIterations 2000]
130 T. Chen et al.

4.4 Results and Analysis


After the processing of Mathmatica, the result of optimization is obtained as
following:
{x1 13.0071,x2 2.47044, x3 60.7104, x4 0.238434, x5 12.9891, x6 13.0407, x7
6.9805, x8 110.067, x9 0.2408021}.
For the nine variables above are practical physical quantities, they are not numbers
continuous; these variables have to be rounded to integer or in accordance with the
standard series. So the result after rounded is
z1 = 13, Z 2 = 71, Z 3 = 155, m1 = 3.5, b1 = 60, x1 = 0.23, i1 = 12.9231,
z1' = 13, z 2' = 15, z3' = 44, m 2 = 7, b 2 = 110, x 2 = 0.24.

Table 1. Parameters Before Optimization

Table 2. Parameters After Optimization


The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 131

15229548.97 14555831.29 (14)


Percentage = 100% = 4.42%
15229548.97

4.5 The Verification and Validation of the Result


In order to assure the effectiveness and reliability of the result we would better
verify the contact fatigue stress and the tooth bending strength fatigue stress using
formula (5) and (6).
(a) Contact fatigue stress:

According to Mechanical Design Handbook, the coefficients above are selected


respectively.
(i) First Stage

With a uniform power source (an electronic motor) and a uniform load (an output
shaft), we find the application factor K A1 = 1.5 ; we have the dynamic
factor KV 1 = 1.1 , and the load distribution factor K 1 = 1.0 ; the valve of load
partition factor K 1 depend on the tooth width of gear b and the pitch diameter d1 .
2
b
K 1 = 0.99 + 0.31 + 0.00012 b = 1.54 (15)
d1
Then, compute the load factor

K1 = K A1 KV 1 K 1 K 1 = 1.5 1.5363 1.0 1.1 = 2.54 (16)

According to Mechanical Design Handbook, we find node region


coefficient Z H 1 = 2.48 ; elastic coefficient Z E1 = 189.98 MPa , contact ratio
coefficient Z 1 = 0.9139 .
So the contact fatigue stress

2 K1T1 (u1 + 1)
1 = Z H 1Z E1Z 1 = 1301.94MPa (17)
b1d12u1

The permitted contact fatigue stress

H lim
HP = Z N ZW (18)
S H min
132 T. Chen et al.

Where: H lim --- endurance limit for contact strength; Z N ---life factor; ZW ---
hardness ratio factor; S H min ---factor of safety.
According to Mechanical Design Handbook, we find endurance limit for contact
strength H lim = 1500 MPa ; life factor Z N = 0.8562 ; hardness ratio
factor ZW = 1.1841 ; factor of safety S H min = 1
Then, compute the permitted contact fatigue stress

H lim
HP = Z N ZW = 1521.11MPa > 1 (19)
S H min
(ii) Second Stage

With a uniform power source (an electronic motor) and a uniform load(an output
shaft),we find the application factor K A 2 = 1.5 ;we have the dynamic factor
KV 2 = 1.1 ,and the load distribution factor K 2 = 1.0 ; the valve of load partition
factor K 2 depend on the tooth width of gear b2 and the pitch diameter d1' .
2
b
K 2 = 0.99 + 0.31 2 + 0.00012 b2 = 1.35 (20)
d'
1
Then, compute the load factor

K 2 = K A 2 KV 2 K 2 K 2 = 1.5 1.35 1.0 1.1 = 2.23 (21)

According to Mechanical Design Handbook, we find node region


coefficient Z H 1 = 2.22 ; elastic coefficient Z E1 = 189.98 MPa , contact ratio
coefficient Z 1 = 0.9491 .
So the contact fatigue stress

2 K 2T2 (u2 + 1)
1 = Z H 2 Z E 2 Z 2
' = 1584.67 MPa (22)
b2 d12' u1

According to Mechanical Design Handbook, we find endurance limit for contact


strength H lim = 1500 MPa ; life factor Z N = 0.9388 ; hardness ratio
factor ZW = 1.1853 ; factor of safety S H min = 1 .
The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 133

Then, compute the permitted contact fatigue stress

H lim
HP = Z N ZW = 1669.14MPa > 1' (23)
S H min
(b) Tooth bending strength fatigue stress

According to Mechanical Design Handbook, the coefficients above are selected


respectively.
(i) First Stage

With a uniform power source (an electronic motor) and a uniform load (an output
shaft), we find the application factor K A1 = 1.5 ; we have the dynamic
factor KV 1 = 1.1 , and the load distribution factor K 1 = 1.0 ; the valve of load
partition factor K 1 depend on the tooth width of gear b and the pitch diameter d1 .
2
b
K 1 = 0.99 + 0.31 + 0.00012 b = 1.54 (24)
d1
Then, compute the load factor

K1 = K A1 KV 1 K 1 K 1 = 1.5 1.5363 1.0 1.1 = 2.54 (25)

According to Mechanical Design Handbook, we can find tooth form


factor YFa1 = 2.32, YFa 2 = 2.29, YSa1 = 1.78, YSa1 = 1.74 , contact ratio
coefficient Y 1 = Y 2 = 1.09 .
So the bending fatigue stress

2 K1T1
F1 = YFa1YSa1Y 1 = 324.60 MPa
b1d1m1
(26)
2 K 2T2
F1' = YFa 2YSa 2Y 2 = 299.69MPa
b2 d1' m2

The permitted contact fatigue stress

F lim
FP = YN YST (27)
S F min
134 T. Chen et al.

Where: F lim --- endurance limit for bending strength; YN ---life factor for bending
strength; YST ---stress correction factor ,usually YST = 2.0 ; S F min ---factor of safety
for bending strength.
According to Mechanical Design Handbook, we find endurance limit for contact
strength F lim1 = F lim 2 = 460MPa ; life factor for bending
strength YN 1 = 0.87, YN 2 = 0.91 ; factor of safety S F min = 1.6 .
Then, compute the permitted bending fatigue stress

F lim1
FP1 = YN 1YST = 499.39MPa > F 1
S F min
(28)
F lim 2
FP 2 = YN 2YST = 522.62MPa > F 2
S F min
(ii) Second Stage

With a uniform power source (an electronic motor) and a uniform load(an output
shaft),we find the application factor K A 2 = 1.5 ;we have the dynamic factor
KV 2 = 1.1 ,and the load distribution factor K 2 = 1.0 ; the valve of load partition
factor K 2 depend on the tooth width of gear b2 and the pitch diameter d1'

2
b
K 2 = 0.99 + 0.31 2 + 0.00012 b2 = 1.35 (29)
d'
1
Then, compute the load factor

K 2 = K A 2 KV 2 K 2 K 2 = 1.5 1.35 1.0 1.1 = 2.2275 (30)

According to Mechanical Design Handbook, we can find tooth form factor


YFa1 = 2.34, YFa 2 = 2.32, YSa1 = 1.81, YSa1 = 1.76 , contact ratio
coefficient Y 1 = Y 2 = 1.29 ..
So the bending fatigue stress

2 K1T1
F1 = YFa1YSa1Y 1 = 416.45MPa
b1d1m1
(31)
2 K 2T2
F1 ' = YFa 2YSa 2Y 2 = 397.31MPa
b2 d1' m2
The Optimization of Two-Stage Planetary Gear Train Based on Mathmatica 135

According to Mechanical Design Handbook, we find endurance limit for contact


strength F lim1 = F lim 2 = 460MPa ; life factor for bending
strength YN 1 = 0.91, YN 2 = 0.94 ; factor of safety S F min = 1.6 .
Then, compute the permitted contact fatigue stress

F lim1
FP1 = YN 1YST = 521.58MPa > F 1
S F min
(32)
F lim 2
FP 2 = YN 2YST = 539.06MPa > F 2
S F min

5 Conclusion

Directed by theory of differential evolution algorithm the optimal design


mathematical model of planetary gear train is established. This mathematical model
with nine design variables and sixteen constraints conditions is a complex optimal
design problem. When simulated in Mathmatica, minimal volume (also minimal
weight) of the planetary gear train can be obtained, and optimal values of variables
can be obtained too.
Then, we verify the optimal result using strength conditions in mechanisms and
machine theory, including the contact fatigue stress and tooth bending strength fatigue
stress. The verification infers that the optimization based on Mathmatica with
differential evolution algorithm is effective and correct. The result of simulation
shows that by using differential evolution algorithm, the weight of the planetary gear
train can be reduced; the design quality and efficiency can be improved greatly.
Differential evolution algorithm is a new method in solving the complex optimal
design problem. The fundamental idea, simulation process and data-processing
method of planetary gear train can be used for reference to other similar optimal
design.

Acknowledgment. This research was supported by the Fundamental Research Funds


for the Central Universities.

References
1. Xue, Y., Lv, G.M., Chen, S.: Optimum design of the planetary g ear box based on
MATLAB. Construction Machinery (5) (2005)
2. Liu, L.M., Li, Y.X.: The optimization of planetary gear reducer based on Mathlab.
Mechanical Engineer. (9) (2009)
3. Sun, Z.L., Li, C.: The research on Multi-objective reliability optimization method of
planetary gear reducer. Machinery & Electronics (10) (2007)
4. Guan, H.J., Zhang, N.H., Liu, B.G.: The optimal design of three stage planetary gear
reducer. Journal of Mechanical Transmission 32(3) (2008)
136 T. Chen et al.

5. Zhu, H.L., Mao, Y., Zhu, B.S., Miao, W.M.: The minimum volume optimal design of
planetary gear reducer in construction machinery and equipment. Construction Machinery
and Equipment 40(11) (2009)
6. Shigley, J.E., Mischke, C.R., Budynas, R.G.: Mechanical Engineering Design, 7th edn.
McGraw-Hill (2004)
7. Mott, R.L.: Machine elements in mechanical design. Prentice Hall, Inc. (2004)
8. Juvinall, R.C., Marshek, K.M.: Fundamentals of Machine Components Design, 3rd edn.
John Wiely & Sons, Inc. (2000)
9. Smith, E.H.: Mechanical Engineers Reference Book, 12th edn. Butterwort-Heinemann,
Oxford (1998)
10. Spoots, M.F., Shoup, T.E.: Design of Machine Elements. Prentice-Hall, Inc. (1992)
11. Edwards Jr., K.S., Mckee, R.B.: Fundamentals of Machine Components Design. McGraw
Hill (1991)
12. Norton, R.L.: Design of Machinery, An introduction to the synthesis and analysis of
mechanisms and machines. McGraw Hill (1999)
13. Oberg, E., et al.: Machinerys Handbook, 25th edn. Industrial Press, New York (1996)
14. Ye, Z., Lan, Z., Smith, M.R.: Mechanism and Machine Theory, Beijing (2001)
15. Eckhardt, H.D.: Kinetic Design of Machines and Mechanisms. McGraw Hill (1998)
16. Erdman, A.G., Sandor, G.N., Kota, S.: Mechanism Design: Analysis and Synthesis.
Prentice-Hall, Inc., New Jersey (1997)
17. Dimarogoneas, A.D.: Machine Design. John Wiley & Sons, Inc. (2001)
18. Mishra, S.K.: The nearest correlation matrix problem: Solution by differential evolution
method of global optimization (2007)
19. http://www1.icsi.berkeley.edu/~storn/code.html

Das könnte Ihnen auch gefallen