modeling of tin coating roughness using fuzzy logic approach.
The coating roughness based on fuzzy logic is realized.
The Insertcutting tool tin is deposited using a physical gas phase (PVD)
Central cube design (CCD)
The optimal experimental points are designed, and the fuzzy rules are developed using the collected experimental data.
In the development of fuzzy rules, the bell-shaped and triangular member functions are proposed. based model.
Results of fuzzy rules-
The validity of the model based on residual error and prediction accuracy is verified.
The residual error of the fuzzy rule model based on the triangular membership function is small, and the prediction accuracy is high, which is 7. 85% and 95.
The results show that fuzzy logic is a good alternative to predict the roughness of TiN coating.
[Key words] Fuzzy logic;
Roughness of coating; PVD;
RF sputtering 1.
The introduction of high-speed machining temperature at the cutting tip may exceed 800oc.
This can result in tool wear, thus reducing tool performance.
Therefore, a tool with high resistance wear is needed to deal with this critical condition.
High wear-resistant tools can directly shorten the tool life and reduce the processing cost.
By coating the film coating on the tool, the coating performance can be improved.
The main purpose of the coating is to enhance the surface properties while maintaining its raised properties.
Compared to the uncoated tool [the coating tool has been shown to be 40 times better in terms of tool wear resistance than the uncoated tool]1].
Hard coatings such as nitrogen and titanium (TiN)
The coating is usually used in the metal cutting industry due to its coating properties such as hardness and wear resistance.
The two main techniques for depositing coatings on knives are physical vapor deposition (PVD)
Chemical vapor deposition (CVD).
The main difference between these two processes is the steam source.
The physical deposition process uses a solid target as the source material and evaporates in atomic particles to form a film coating.
However, the CVD process uses chemical sources as coating materials.
In the cvd coating process, the sputtering particles embedded in the hard material on the tool can lead to the presence of the reaction gas.
A process in physical deposition technology, that is, magnetic absorption and sputtering is a good process.
In the hard coating industry, the well-known technology, it is able to spray many hard materials such as titanium coated knives.
It is reported that in the cvd coating process, many factors have a significant impact on the coating properties, including the coating roughness [2-4].
Surface roughness is one of the important features that affect the machining performance.
It affects the level of friction and material pickup
Up behavior of upward sliding of tool and workpiece material .
Some studies have shown that n2 pressure, nitrogen pressure and turntable speed have a significant effect on the roughness and surface morphology of the deposited surface [6-8].
Modeling is an effective way to solve cost and customization problems in painting process.
It can be used to predict the value of the coating properties and find the best parameters in many processes.
Many methods in experimental and intelligent design
The technology-based modeling method has been applied to Tian Kou full-factor neural network and fuzzy logic .
The design of experimental methods is usually used to collect important experimental data .
However, some limitations of these methods have been discussed.
The Tian Kou method is difficult to detect the interaction effect of the nonlinear process, and the full factor is only suitable for optimization purposes [12 13].
Neural networks require a large amount of training data to be robust .
The Fuzzylogic method builds rules using actual data or expert advice.
This technique has been used to model uncertain complex problems. 15].
Therefore, the application of fuzzy rules
This paper discusses a method based on prediction of surface roughness of TiN coating.
Table 1: setup of the physical gas phase deposition reaction sputtering process 1 Materials and methods experiments were carried out in the unbalanced physical gas phase deposition reaction system manufactured by VACTEC South Korean model VTC physical gas phase deposition 1000.
Titanium installed vertically (Ti)target.
Wash the surface of the carbide blade in an ultrasonic cleaning machine with an alcohol bath for 20 minutes.
The carbide blade is loaded in a heated substrate bracket at the interior part of the coating.
During the sputtering process, an inert gas called ar was used in the coating chamber to generate electrons.
In the presence of nitrogen, the tool blade is coated with Ti.
Table 1 shows the detailed setup of the coating process.
During this process, N2 pressure ar pressure and turntable speed were selected as variables Table 2: language variables and ranges for triangle and Bell member Table 3: rules built based on actual data Table 4: residual errors and accuracy of fuzzy models using triangular member Table 5: residual errors and accuracy of fuzzy models using Bell membership 2.
2 Experimental design in this study is based on the experimental matrix of central cubic design (CCD)
Use design expert version 8. 0 software.
It is designed based on 8 Analysis points, 6 axis points and 3 center points.
Operation window)were set as +/-
Alphavalue, and based on that point, the software assigns high settings and low settings for the cause point.
The role of the pole is to ensure representation in a wide range of operating windows. 2.
3 The surface roughness value of the atomic force microscope of the tin coating was examined by scanning force microscope (AFM)method.
This method determines the morphology of the surface based on less sample preparation and non-sample preparation requirements
Destructive TestThe AFM XE-
The 100 model operates at room temperature. Non-
The scanning area is set to 25A-using the commercial cantilever contact mode detection method-25microns (625 m2).
Then the surface image is analyzed by XEI software to get the surface roughness reading. 3.
0 model fuzzy logic based on fuzzy rules is determined as a set of mathematical principles based on membership knowledge representation, rather than a clear membership on classical binary logic 16].
This theory is developed on the basis of fuzzy set theory.
It involves the degree of membership and the degree of truth in solving multiple problems.
The position of the value, not just the black and white that changes continuously from zero (nota member)to one (
Definitely a member).
For example, in the measurement of surface roughness, the precise mathematical meaning (e. g. 44. 0 m)
Subjective language terms can be used (e. g. low)
Through fuzzy set theory.
On the contrary, language terms can also be defined with precise mathematical meaning.
Typical steps to develop fuzzy rules
The model-based implementation includes the following steps: 1.
Specify the problem and define the language variable. 2.
Determine the fuzzy set. 3.
Extraction and construction of fuzzy rules. 4.
The fuzzy rules and processes of fuzzy sets are encoded.
Evaluate and validate the model.
Step 1: specify the problem and define the language variable.
In this study, the three input variables for the sputtering process were N2 pressure, ar pressure and turntable speed, and the output response was surface roughness.
All variables and responses are called language variables.
The language variables were adjusted according to the range of experimental parameters.
For example, N2 pressure is separated in the range of 0. 16-1. 84A-10-3 mbar.
Each language variable is then separated into language values.
The input variable uses three low language values (L)Medium\' (M)andHigh\' (H).
In the range of 40, the output variables are also divided into three language values. 0-100.
0 m, any value is lower than this range until 0 is considered low and higher than this range is considered high.
Step 2: determine the fuzzy set.
The model uses triangle and bell-shaped membership functions to describe fuzzy sets of input variables and output responses.
In many applications, both of these membership functions are used to develop models based on fuzzy rules [17 18].
The triangle and Bell functions depend on three parameters, such (1)and (2)respectively.
Table 2 shows the language variables and ranges of the ship embership function.
Step 3: Extraction and construction of fuzzy rules.
The construction of fuzzy rules is a difficult point in this study.
Seek advice from experts at this stage.
Fuzzy rules can be defined as conditional statement equations where language variables and B and C are language values determined by fuzzy sets in the discourse X Y and Z universes, respectively .
And is the input And output of the rule.
The value depends on the combination of the usingAND operator.
In practice, fuzzy rules are developed based on expert suggestions for variable combinations and output responses.
However, due to the non-linearity of the manufacturing process, the construction of the rules may not be accurate.
In this work, fuzzy rules are developed based on collected experimental data.
A set of 15 rules are constructed that can be defined as shown in step 4 of Table 3: encoding fuzzy rules and procedures for fuzzy sets to execute the model.
At this stage, the fuzzy set fuzzy rules and programs are written into Matlab programs.
The following code is an example of a program command that uses the Bell Model\'s input N2 pressure variable.
Step 5: Evaluate and validate the model.
Many types of performance metrics can be used to evaluate the predictive performance of rulesbased model.
The remaining error is as follows (3)
Is a method to quantify the difference between the predicted value of the predicted quantity and the actual value.
Model prediction accuracy as shown in ()4)
Calculate to determine the accuracy of the model. Equation 4.
0 results of validation and discussion three test data sets from the separation experiment were used to verify the fuzzy rulesbased model.
Table 4 shows the maximum residual error of fuzzy rules-
The model based on the triangular membership function is 7. 85%.
The table also shows that the prediction accuracy of the developed model is 95. 05%.
At the same time, as shown in Table 5, the maximum residual error and prediction accuracy of fuzzy rules
The base model using the Bell membership function is 9. 24% and 90. 20% respectively.
Small residual error of two fuzzy rules-
Basedmodels shows that the prediction results are very close to the actual experimental values.
In addition, the results of high precision show that fuzzy rules
The prediction of TiNcoatings roughness based on the model has good prediction performance.
Fuzzy rules in comparison-
Compared with the model using the bell-shaped membership function, the prediction accuracy based on the triangular membership function is higher. 5.
0 conclusion in this study, the prediction of tin coating was obtained using fuzzy rulesbased models.
Based on the collected experimental data, fuzzy rules are constructed.
The input parameters are nitrogen pressure, nitrogen pressure and turntable speed, and the output response is TiN coating roughness.
Select the triangle and bell shape as the membership function of the input and output fuzzy set. The fuzzy rule-
The model is verified with three experimental data.
Results were observed based on residual errors and model accuracy.
The results show that even in a small amount of data, the data collected by CCD technology can be used in the development of fuzzy rules.
Maximum residual error of fuzzy rules-
The membership functions based on triangles and bell shapes are 7, respectively. 85% and 9.
The minimum value of the error shows that the prediction result is very close to the actual experimental value.
The fuzzy rule model based on triangular MFs shows 95.
Prediction Accuracy 05%.
At the same time, the model with bell-shaped MFs is displayed as 90.
Prediction Accuracy 20%.
The results show that the model has good prediction performance.
Therefore, fuzzy logic is a good choice to predict the surface roughness of TiN coating.
The author thanks the University of Malacca, Malaysia (UTeM)for grant no. PJP/2013/FTMK(14A)/S01218. REFERENCES K. Tuffy G. Byrne and D.
\"Determination of the best tin coating thickness on WC blades for processing carbon steel\" Journal of Material Processing Technology Volume 1155-156 pp. 1861-1866 2004. P. H. Mayrhofer C. Mitterer L. Hultman and H.
Clement\'s material science progress volume for the microstructure design of hard coatings51 pp. 1032-1114 2006. J. Musil and J.
Vleck \"control of the texture and particle size of the film\" material chemistry and physics. 54 pp. 116-122 1998. D. L.
Smith Film Deposition: Principle and Practice.
New York: McGraw Hill 1995B. Podgornik S. Hogmark and O.
Sandberg surface roughness and coating type impact on wear performance of coated forming tool steel surface and coating technology roll184 pp. 338-348 2004. K. Chakrabarti J. J. Jeong S. K. Hwang Y. C. Yoo andC. M.
Effect of Lee \"nitrogen flow rate on growth morphology of tiain film prepared by rf-
Solid film roll of reactive sputtering technology406 pp. 159-163 2002. H. C. Jiang W. L. Zhang W. X. Zhang and B.
Effect of Argon \"ar pressure on magnetic properties and low magnetic properties
Field giant effect of amorphous TbFe film405 pp. 834-838 2010. M. Matsumoto K. Wada N. Yamaguchi T. Kato and H.
Matsutter \"effects of substrate rotation speed during deposition on thermal cycling life of thermal insulation coating prepared by electron beam physical vapor deposition\" surface and coating technology roll202 pp. 3507-3512 2008. H. Cetinel H. Ozturk E. Celik and B.
Artificial neural network-Karlik
Prediction technology based on wear volume in Mocoating \"wear volume\"26 pp. 1064-1068 2006. R. Manaila A. Devenyi D. Biro L. David P. B. Barna and A.
Surface and coating technology of Kovacs \"multi-layer tiain coating with component gradient\"151-152 pp. 21-25 2002. F. Karacan U. Ozden and S.
Karacan \"Optimization of manufacturing conditions of Turkish Brown activated carbon by chemical activation using response surface method\" application thermal engineering roll27 pp. 1212-1218 2007. M. J. Anderson andP. J.
Simplification: practical tools for effective experiments.
Portland OR: productivity publishing house 2000. S. Bisgaard and N. T.
Diamond \"analysis of the test method of Tian Kou confirmation\" 1990. F. A. N.
Fernandez and L. M. F.
Lona, Brazilian Journal of Chemical Engineering, \"neural network application in aggregation process\"22 pp. 401-418 2005. M. D. Jean B. T. Lin and J. -H.
Zhou \"design of fuzzy logic method for optimizing enhanced zirconium deposition using plasma spraying\" surface and coating technology roll201 pp. 3129-3138 2006. L. A.
Information and Control of \"fuzzy set\" in Zadeh. 8 pp. 338-353 1965. S. Yagiz and C.
Gokceoglu \"application of fuzzy reasoning system and nonlinear regression model in rock brittle prediction\"37 pp. 2265-22722010. D. Chatterjee G.
Sutradhar and B.
Fuzzy rules of Oraon
Journal of Materials Processing Technology hardness prediction of sintering steel components based on hardness200 pp. 212-220 2008. M.
Artificial Intelligence: A Guide to Intelligent Systems. Essex England. : Addition-Wesley 2005.