Machine vision based monitoring system for milling using ANFIS · Machine vision based monitoring system for milling using ANFIS 1John Stephen R, 2Palani S ,3Alagu sundara pandian

Machine vision based monitoring system for milling using ANFIS

1John Stephen R,

2Palani S,

3Alagu sundara pandian,

4Paul Linga Prakash R and

5Vijayakumar

D

1,2,34,5Department of Mechanical Engineering, VEL TECH MULTITECH, Avadi, Chenna-62, India

Corresponding Author E-mail ID: [email protected]

Abstract

The surface roughness of machined work

pieces are the important discussion since the

fitness of the product mainly related to the

roughness of the machined surface. In our

research work deals with the rough

monitoring by machine vision technique on

the machined surface. Here, the machined

surface images are captured by vision

system. A novel adaptive neuro fuzzy

inference system (ANFIS) model is used to

correlate the prediction values with

experimental value of the input and the

response parameters in the end milling

operation. The ANFIS results shows error

while extraction of image values feed as

ANFIS training process. Correlation

accuracy encouraging us for suitable

monitoring technique of the proposed ANFIS

model with image extraction technique with

image processing method and it is very

suitable for online monitoring in the

automated production industries.

Keyword. Milling, machine vision, Non

contact, surface roughness;

ANFIS

1. Intriduction

In the modern manufacturing era, end

milling process is essential one as well as

regularly utilizing for machining of the

metal. Recently, manufacturing components

quality as main concern in the competitive

industrial sectors and needed to online

monitoring methods [1]. The work piece

surface roughness could be measured

through two ways such as contact as well as

non-contact methods. In the method of

contact sensor probe has touch on the

machined surface lead to form scratches on

it. In order to avoid physical touch during

measurement the on line image processing

technique as well as machine vision methods

are introducing in this research work [2].

In computerized manufacturing scenario,

computer machine vision method is a main

play role for fast monitoring [3]. The regular

research has been followed through

computer machine vision method into the

modern manufacturing sector., The non-

contact online monitoring methods very

suitable for machined surface roughness

measurement without affecting the already

machined surface.[4].

Recent day’s machine vision based

techniques eliminated disadvantages of the

conventional probe measuring methods.

Galante et al stated that the vision approach

International Journal of Pure and Applied MathematicsVolume 119 No. 15 2018, 1083-1090ISSN: 1314-3395 (on-line version)url: http://www.acadpubl.eu/hub/Special Issue http://www.acadpubl.eu/hub/

1083

in the process of turning for prediction [5].

Benardos used taquchi's optimization design

method for prediction in the process of face

milling [6]. Lee constructed a model to

measure the roughness through online

monitoring for the operation of turning [7].

In our research work we used neuro fuzzy

logic for correlating actual as well as

predicted roughness values on the work

piece with various condition of machining.

In our research work firstly explain machined surface image extraction and then ANFIS method has been introduced. After that shows the ANFIS prediction results for the different machining conditions with the results of the correlating values with the percentage error. Finally our research work take part of result as well as discussion with model output and conclude the work for possible experimental approaches with accurate dimensions for optimizing the machining time on end milling process.

2. Measurement of roughness of the

work piece

In 1960 the image processing method

was introducing especially for the

investigation of space for generating graph

of earth. Now a day, it improved for

computer aided manufacturing as well as all

sector of developed industrial area for

monitoring by machine vision [8]. The

purpose of vision technique for online

monitoring investigated through the

literature [9]. Bradly [10] developed a novel

online measurement and monitoring model

for roughness through image processing

method.

In our research work, image and vision

method has been used for predicting

roughness of the surface in end milling as

shown in Figure 1. First sources of light has

been focus on job next image of the

machined surface has been accrued through

vision system then directed to work station

with personal computer through frame

grabber. WE use a Surf coder for measuring

the machined surface roughness of work

material with the dimension of 8 mm long

and the speed is 0.5 mm/s. The mean of

arithmetic roughness of the machined

surface Ra has been measured through the

formulae below:

(1)

Where Yi is the height of roughness from its

mean

n is the experimental date number Ga is the average gray value of the

image of work piece machined surface.

International Journal of Pure and Applied Mathematics Special Issue

1084

Figure 1. Rapid I of Computer vision

model

V: 28 m/min V: 28

m/min

F: 0.002 mm/rev F: 0.0466

mm/rev.

D: 0.6 mm D: 0.8

mm

R: 0.6371 µm R: 0.8517 µm

G: 109.563 G: 153.845

Figure 2. Captured work piece Images

The expression of the gray value in

arithmetic average is shown in equation 2:

(2)

Where Gi is the average gray value of the

image of the surface

3. ANFIS Model for surface rough

Problem has been generating through

fuzzy rules with membership function.

These difficulties are overcome with ANFIS

through explicit adaptive knowledge of

fuzzy with implicit learning capacity of

neural networks [11]. The Artificial neural

network (ANN) gives experts neurons to

generate fuzzy control if then rules including

membership for obtaining suitable

formulations methods [12].

The advantages of both traditionally

fuzzy expert and the neural networks

combined with the practice of computer to

obtain computational performance by layer

of hidden neurons as well as the learning

capacity. The model structure of ANFIS is

shown in Figure 3.

Figure 3: Model structure of ANFIS

4. Experimental discussion

The practical status of CNC milling

process is shown in Figure 4


1085

Figure 4: Machining status of the CNC

milling

Experimental Results

Table 1: End milling process experiment

values

Sl.

No

V

m/min

F

mm/rev

D

mm Ga

Ra

microns

1 28 0.002 0.6 0.6607 0.6371

2 28 0.002 0.8 0.8498 0.5705

3 28 0.002 1.0 0.7506 0.5039

4 28 0.0333 0.6 0.8420 0.777

5 28 0.0333 0.8 0.6621 0.67

6 28 0.0333 1.0 0.6679 0.644

7 28 0.0466 0.6 1.0000 0.9183

8 28 0.0466 0.8 0.9278 0.8517

9 28 0.0466 1.0 0.7445 0.687

10 38 0.002 0.6 0.8651 0.37

11 38 0.002 0.8 0.7073 0.3025

12 38 0.002 1.0 0.8339 0.2359

13 38 0.0333 0.6 0.7592 0.5097

14 38 0.0333 0.8 0.9069 0.4431

15 38 0.0333 1.0 0.8803 0.3765

16 38 0.0466 0.6 0.6744 0.6503

17 38 0.0466 0.8 0.8694 0.5837

18 38 0.0466 1.0 0.8743 0.587

19 47 0.002 0.6 0.8097 0.0827

20 47 0.002 0.8 0.1574 0.0469

21 47 0.002 1.0 0.1460 0.0321

22 47 0.0333 0.6 0 .8543 0.2417

23 47 0.0333 0.8 0.8342 0.236

24 47 0.0333 1.0 0.4935 0.01321

25 47 0.0466 0.6 0.8938 0.3823

26 47 0.0466 0.8 0.7381 0.3157

27 47 0.0466 1.0 0.7754 0.286

The experimental values are tabulated in

Table 1. 27 experiments through different

milling conditions have been obtained to

train the ANFIS for monitoring roughness of

the work piece surface. The experimental

values were tabulated in Table 1.

5. Prediction of Machined surface

Input data Normalisation

Normalization technique has been utilized

for obtaining test values from -1 to +1. The

normalization process mainly used for

training the parameter values uniformly as

well as to meet equal gap to the training

parameters to increase performance of

ANFIS method at the same time reducing


1086

the computation time. Figure 5 shows the

ANFIS training.

Figure 5: ANFIS training

For training ANFIS first proper topology

has been chosen through trial-error method.

Then various topologies have been tried and

4-4-1 topology network has been selected

for training the ANFIS. Figure 5 shows the

created ANFIS.

Finally, trained ANFIS has been

simulated through new input for testing to

predict roughness of the machined surface.

Input data for valitation

Five new sets of data have been taken to

validate proposed ANFIS model. Table 2

shows the values for input as well as output

variables. The % of error in the predicted as

well as experimental value have been

obtained through following formula.

Predicted Value - Experimental value

% of Error = ------------------------------------------

×100

Experimental value

Table 2: Validation between predicted with

experimental values

Sl.

No

V

m/m

in

F

mm/

rev

D

mm Ga

Ra in microns Error

%

Experi Predic.

1 1 0.0429 0.6 0.809

7 0.0511 0.0493 3.523

2 1 0.0429 1.0 0.146 0.0349 0.0311 10.888

3 1 0.7146 1.0 0.493

5 0.0132 0.0130 1.665

4 1 0.0429 0.6 0.809

7 0.0901 0.0832 7.658

5 1 0.7146 0.6 0.854

4 0.2632 0.2599 1.253

Average Absolute Error = 4.997 %

The output of the validation is shown in

Figure 6.

Figure 6: Output of the validation by

ANFIS


1087

6. Result and discussion

Proposed model has been developed

through ANFIS. With twenty seven various

machining parameters are as shown in Table

1 used to predict the roughness of the

machined surface in the end milling. The

validation of the predicted values with

experimental values is shown in Figure 7.

Figure 7: Validation of experimental with

ANFIS predicted data

The mean error of percentage of the

proposed model with ANFIS is 4.997% and

accuracy is 95.003%.

7. Conclusion

The results shows that an on line

measuring as well as monitoring method by

machine vision method for investigating

machined surface of the roughness of work

piece by ANFIS. Through this results this

proposed ANFIS model as well as image

processing technique is best suited for

monitoring methods. The outputs of

experiment are comparing through results of

the proposed method shows the proposed

model could be produced very accurate

result.

REFERENCES

[1] G.A. A1. Kindi, R.M., K.F. Gill, “An

Application of machine vision in the

automated inspection of engineering

surfaces” International Journal of

Production Research Vol.30,

No.2,pp,241-253,1992.

[2] M.B. Kiran, B, Ramamoorthy,

B.Radhakrishnan, “ Evaluation of

surface roughness by vision system “,

International Journal roughness by

vision system”, International Journal of

machine Tools & Manufacture Vol.38,

No 5-6, pp 685-690, 1998

[3] M.Gupta, S. Raman, “Machine vision

assisted characterization of machined

surfaces”, International Journal of

Production Research Vol.39,No pp.759-

784,2001.

[4] K. Venkata Ramana, B. Ramamoorthy,

“Statistical methods of compare the

texture features of machine surfaces”,

Pattern Recognition. Vol29, No9 pp,

1447-1459, 1996

[5] Galant G, piacentini M. Ruisi VF.

Surface roughness detection by tool

image processing Wear 1991;148:211-

20.


1088

[6] P.G.Benardos, G.C. Vosniakos,

Prediction of surface roughness in CNC

face milling using neural networks and

Taguchi’s design of experiments,

Robotics and Computer Integrated

manufacturing 18 (5-6) (2002) 343-354

[7] B.Y.Lee, S.F.Yu, H.Juan “The model of

surface roughness inspection by

machine vision system in turning,

Mechatronics 14 (2004) 129-141.

[8] Dudas I and Varga G 2001 Metrological

use of CCD Camera microCAD 2001

Int. Scientific Conference March 1-2

Univ. of Miskole, Hungary 35-40

[9] Kurada S Bradley C (1997) A review of

machine vision sensors for tool condition

monitoring, Comput Ind 34(1)55-72

[10] Bradly C Wong YS (2001) surface

tecture indicators of tool wear – a

machine vision approach. Int. J Adv

Manuf Technol 17(6):435-443

[11] Stamotios V. Kartalopoulos, (1996)

“Understanding Network and Fuzzy

Logic” Prentice-Hall of India Ltd, New

Delhi

[12] Jang, J.S. Sun, C.T. Mezuzah,E. (1997)

“Neuro Fuzzy Logic and Soft

Computing”. Prentice-Hall Englewood

cliffs


1089

1090

Documents

Machine vision based monitoring system for milling using ANFIS · Machine vision based monitoring system for milling using ANFIS 1John Stephen R, 2Palani S ,3Alagu sundara pandian