Modelling of Thrust Forces in Drilling of AISI 316 Stainless Steel Using Artificial Neural Network and Multiple Regression Analysis
Tam metin
(2) 1 86 7 2 2,* 8= 3 1 Z8 D
(3)
(4) 1 =6 6 2 Z8 D
(5)
(6) <
(7)
(8) =R
(9) I
(10) 6 3 7 D
(11)
(12) 1 =I
(13)
(14) 6. In this study, the effects of cutting parameters (i.e., cutting speed, feed rate) and deep cryogenic treatment on thrust force (Ff) have been investigated in the drilling of AISI 316 stainless steel. To observe the effects of deep cryogenic treatment on thrust forces, M35 HSS twist drills were cryogenically treated at –196 °C for 24 h and tempered at 200 °C for 2 h after conventional heat treatment. The experimental results showed that the lowest thrust forces were measured with the cryogenically treated and tempered drills. In addition, artificial neural networks (ANNs) and multiple regression analysis were used to model the thrust force. The scaled conjugate gradient (SCG) learning algorithm with the logistic sigmoid transfer function was used to train and test the ANNs. The ANN results showed that the SCG learning algorithm with five neurons in the hidden layer produced the coefficient of determinations (R2) of 0.999907 and 0.999871 for the training and testing data, respectively. In addition, the root mean square error (RMSE) was 0.00769 and 0.009066, and the mean error percentage (MEP) was 0.725947 and 0.930127 for the training and testing data, respectively. Keywords: artificial neural networks, regression analysis, cryogenic treatment, machining, thrust force, predictive modelling. 0 INTRODUCTION ;;
(15)
(16) ==
(17) 2
(18)
(19) ;
(20)
(21) H ;
(22) =
(23)
(24) ; = ==
(25)
(26) ;
(27) =
(28)
(29)
(30) 3 &H <
(31)
(32) ;; ;
(33)
(34) ; ; ==
(35)
(36)
(37)
(38) / 4H
(39)
(40)
(41)
(42)
(43) =
(44) ;
(45)
(46)
(47)
(48)
(49)
(50)
(51) =
(52) =
(53) =
(54)
(55)
(56)
(57) H 6
(58)
(59)
(60) =
(61)
(62)
(63) =
(64) = R
(65) H 6
(66) ;
(67)
(68) = = 4 &/&³ ;
(69)
(70)
(71)
(72) 0H =
(73) ; =
(74) =
(75)
(76) ;
(77) H ;
(78) =
(79) ;
(80)
(81)
(82)
(83) ;
(84) H Z;
(85) ;
(86) =
(87)
(88)
(89) =
(90) =
(91)
(92) ;.
(93) ; ;
(94)
(95) ; H =;
(96) =
(97) =
(98)
(99)
(100) ; = ; =
(101)
(102)
(103)
(104)
(105) ;
(106)
(107)
(108)
(109) =
(110) ;
(111) ; H 6= ;
(112) ;;
(113) =
(114)
(115)
(116) ==
(117) 'H 492.
(118) =
(119)
(120)
(121) ;
(122) =
(123)
(124)
(125) ;=
(126)
(127) = HR H
(128) ;
(129)
(130)
(131) =
(132) =
(133)
(134) *[[+H 6
(135) ; [[
(136) =
(137)
(138) = ; =
(139)
(140)
(141)
(142) ; =
(143) =
(144)
(145) = H 6 ;= = ;
(146) = =
(147)
(148) = H Ó8 7 ; ! = =
(149)
(150)
(151) [[
(152) H$
(153)
(154) = ;
(155)
(156) = = H$ H; =88 ;;
(157) =;
(158)
(159) =
(160) H9 ;
(161)
(162) =
(163) = =88 ;;
(164) . H
(165) 3[[.
(166) ;
(167) = =!/!H6[[
(168) =
(169)
(170) ;
(171)
(172) ;
(173) H [ H &
(174)
(175) == =
(176)
(177)
(178)
(179) ; = = H 1
(180) ;; = 2;
(181)
(182) ;=H I2;
(183) [[ ;
(184)
(185) ;
(186)
(187)
(188)
(189) H R = ;
(190) [[
(191)
(192) . ; H"
(193) / ;
(194) . ;
(195)
(196) = =
(197) = H6
(198) = =. *Corr. Author’s Address: Düzce University, Cumayeri Vocational School of Higher Education, 81700, Cumayeri, Düzce, Turkey, gurcansamtas@duzce.edu.tr.
(199) Strojniški vestnik - Journal of Mechanical Engineering 58(2012)7-8, 492-498.
(200)
(201)
(202) ; ;
(203)
(204) H 6 ;
(205) =
(206)
(207)
(208) . ;
(209)
(210) ;
(211)
(212)
(213) H 6#
(214) =
(215)
(216) == =
(217) ==
(218) ; =
(219)
(220)
(221) =&
(222) ;
(223) =
(224)
(225) ;
(226)
(227) [[
(228) . = ;2
(229)
(230) 2;
(231)
(232) H 1 EXPERIMENTAL DETAIL
(233) &
(234)
(235)
(236) ;
(237)
(238) H6
(239)
(240) =
(241) ;
(242)
(243) !!µ0!µ4H6
(244) ;
(245)
(246) =&
(247)
(248)
(249)
(250)
(251)
(252) 6 H Table 1. The chemical composition of AISI 316 stainless steel C 0.05. Si Mn P S Cr 0.380 0.971 0.039 0.006 16.58. Ni 9.94. Mo Cu 2.156 0.321. 6
(253)
(254) ;=
(255) W = ><'4!<[<
(256)
(257) =
(258) ;*V3/ '
(259) + = *f!H!'!H !H3+H?
(260)
(261) ; = *& +H 2;
(262) >&4 R *Z[ '0+
(263)
(264) H 6
(265) =
(266)
(267)
(268)
(269)
(270)
(271) H . =
(272) *
(273) +
(274)
(275) =
(276)
(277) == = ;
(278) H 6 ; =
(279) *Ct+ %
(280)
(281) *<R6+
(282)
(283) *<6+.
(284) 3 ;
(285) *<66+ 3!!Ð<H6
(286) =>&4 R
(287) ;=
(288) ; = ; Á Ð< ;;2
(289) H4 Ð<
(290) H 6
(291)
(292)
(293)
(294) ; = 3/
(295) ; H ?= 2;
(296)
(297)
(298)
(299) == = = = ;
(300) H Z
(301)
(302)
(303) = 7
(304) 340? ; H & 2;
(305) ;=
(306)
(307) =
(308) ; = ;;
(309)
(310) H.
(311)
(312) [[ = ;;
(313)
(314)
(315)
(316)
(317) H [[
(318) =
(319)
(320) =
(321) =
(322) =
(323)
(324)
(325)
(326) ;2
(327)
(328)
(329)
(330)
(331) ; ; ; 4 0H [[
(332) =
(333)
(334) ;
(335) ; H 6
(336)
(337) ; = =2
(338)
(339) H
(340) ; ;
(341) =
(342) ;
(343)
(344)
(345)
(346)
(347) =
(348) H6
(349) =
(350)
(351) ;=
(352) I H% n. NETi = ∑ wij x j + wbi , . *+. j =1. NETi
(353)
(354) =
(355) ; i ;
(356) wij
(357)
(358) = .
(359) i j ;
(360) xj
(361) ;=j;
(362) . wbi
(363)
(364) =
(365) H 6
(366)
(367) =
(368) ;
(369)
(370)
(371)
(372) ; ; H
(373)
(374)
(375)
(376) ; = ;
(377) .
(378) ; H6
(379) = ;;;
(380)
(381)
(382) =
(383)
(384)
(385) =
(386) == ;= H 9
(387)
(388)
(389)
(390) = =
(391)
(392)
(393) =
(394)
(395)
(396) ; ;
(397)
(398)
(399)
(400) ==
(401)
(402)
(403) =
(404) H 1
(405)
(406)
(407)
(408)
(409) ==
(410)
(411)
(412) =
(413)
(414)
(415) H6
(416) =
(417) ; ! = =
(418) ;H6
(419)
(420)
(421)
(422) =
(423)
(424)
(425)
(426)
(427) I H3H. f ( NETi ) =. 1 . 1 + e − NETi. *3+. 6
(428)
(429)
(430) = [[ ;;
(431) & 2;
(432) = H 1
(433)
(434) &
(435)
(436)
(437)
(438) 30
(439)
(440) H6= ;
(441)
(442) ;=
(443)
(444) ;H
(445) ; 2;
(446)
(447)
(448) ;
(449) =
(450)
(451)
(452) H 1
(453) < ">
(454)
(455)
(456) == . Modelling of Thrust Forces in Drilling of AISI 316 Stainless Steel Using Artificial Neural Network and Multiple Regression Analysis. 493.
(457) Strojniški vestnik - Journal of Mechanical Engineering 58(2012)7-8, 492-498. =
(458)
(459)
(460)
(461) = = H =
(462)
(463)
(464) .
(465) =;
(466)
(467) == <%&4;
(468) *1
(469) H+H. ;
(470) =[[
(471)
(472) ;
(473)
(474) =2;
(475) H9>I* + 92 *
(476)
(477) ==
(478)
(479) +. >I$ * ; + = ;
(480) H6
(481) =
(482)
(483) % ⎛⎛ 1 ⎞ RMSE = ⎜ ⎜ ⎟ ∑ t j − o j ⎜ p j ⎝⎝ ⎠. (. ⎛ ∑ j t j − oj R = 1− ⎜ 2 ⎜⎜ j oj ⎝ ∑ 2. MEP ( % ) =. Fig. 1. ANNs architecture with a single hidden layer.
(484)
(485)
(486)
(487)
(488) = =
(489)
(490)
(491) H =
(492)
(493) =
(494) =
(495)
(496) ! ; H 6
(497) ;. ;
(498) =
(499)
(500)
(501) ; H
(502)
(503) ;. ;
(504) !H !H
(505) ;
(506) ;
(507)
(508) H 6
(509)
(510) ; =
(511)
(512)
(513)
(514) ! 4! ;
(515) H 1
(516)
(517) ;
(518) /H 6
(519)
(520)
(521) =[[
(522) <R6´<6´3 <66´&
(523) .
(524) H [[
(525)
(526) ;
(527)
(528) = . ( ). ). 1/ 2. 2. 2. *&+. ⎞ ⎟, ⎟⎟ ⎠. */+. ⎛ t j − oj ⎞ ⎟⎟ × 100 ⎝ tj ⎠ , p. ∑ j ⎜⎜. *4+. t
(529) o
(530) ; p
(531) = ;H 2 DISCUSSION AND RESULTS 2.1 The Effects of Heat Treatments and Cutting Parameters on the Thrust Forces 6
(532)
(533)
(534) = ;
(535) =
(536) ;
(537)
(538) 1
(539) H3H
(540) 1
(541) H3=
(542)
(543)
(544)
(545)
(546) =
(547) H =
(548) ==
(549) ;
(550)
(551) . Fig. 2. The variations in thrust force depending on the heat treatment, feed rate and cutting speed. 494. ⎞ ⎟⎟ , ⎠. {| }~` }
(552) }{.
(553) Strojniški vestnik - Journal of Mechanical Engineering 58(2012)7-8, 492-498.
(554) H
(555)
(556)
(557)
(558) ; = =
(559) ==
(560)
(561)
(562)
(563) =
(564)
(565) ;H6
(566)
(567) =
(568) == = ;
(569)
(570)
(571) ;H 6 = = <6 <66
(572) <R6
(573) H 6
(574)
(575) ; =
(576) ;
(577)
(578) =
(579) =<6 <66
(580)
(581) ;
(582)
(583) <R6
(584)
(585)
(586)
(587) ; 'H Table 2. Experimental results Cutting speed. Feed rate. 12 14 16 18 12 14 16 18 12 14 16 18. 0.08 0.08 0.08 0.08 0.1 0.1 0.1 0.1 0.12 0.12 0.12 0.12. CHT 1 Thrust force 680 662 660 668 770 765 756 767 894 871 867 870. Cutting tools CT 2 Thrust force 662 645 640 635 765 757 752 761 855 847 844 848. CTT 3 Thrust force 641 634 628 632 757 742 737 749 846 836 839 842. 6
(588) =
(589)
(590)
(591) =
(592) ;
(593) =
(594) = ;
(595) ;
(596) ;
(597)
(598) =
(599)
(600) ; <66
(601)
(602) = 3H6 = ;
(603) [[
(604) ;
(605)
(606)
(607) ==
(608)
(609)
(610)
(611)
(612)
(613) H 6 =
(614) 2;
(615)
(616)
(617) 6 3H6== =
(618) ==
(619)
(620) ; = =
(621)
(622)
(623) 2;
(624) [*
(625) =
(626) +H. 2.2 Prediction of Thrust Forces with Multiple Regression Analysis 9
(627)
(628)
(629)
(630)
(631) =
(632)
(633)
(634) ;=
(635) ;
(636) *x1, x2 ,…, xk+
(637)
(638) ;
(639) *y+H 6
(640)
(641)
(642) =
(643) ;
(644) . = 2 ;
(645) ;
(646) H
(647)
(648)
(649)
(650) [ ;
(651) =
(652) ;
(653) .
(654) H 6 ;
(655) = = [ ;
(656)
(657)
(658) =
(659) ;
(660)
(661) H
(662) ; ==
(663)
(664)
(665)
(666)
(667) ;
(668)
(669)
(670)
(671) ;
(672)
(673) H6
(674) =
(675)
(676)
(677) ; = == =
(678)
(679) 2;
(680)
(681)
(682) H 6
(683) [
(684)
(685) 6 &H
(686) 6 & =
(687)
(688)
(689) =
(690) = ==
(691) = *0H!'³+H 6
(692) =
(693) =
(694) *H³+.
(695) ; *!H3&³+H 6 2;
(696)
(697)
(698) =
(699)
(700) =
(701) = 4³ * 4³ =
(702) +.
(703)
(704)
(705) 1
(706) H &H 6
(707)
(708)
(709)
(710) ;
(711)
(712) = =
(713)
(714) I H*+H6 ==
(715)
(716) =
(717)
(718) *R2+ =
(719) !H3% Ff = 294.861 − 14.458Ct − 1.833V + 5150 f , *+ Ff
(720) = [Ct
(721)
(722) V
(723)
(724) ;
(725) f
(726) = H
(727)
(728)
(729)
(730)
(731)
(732)
(733) ;
(734)
(735) ; =
(736) = H. Table 3. Analysis of variance for thrust force Source Model Cutting Tools Cutting Speed Feed rate Error Total. Degrees of freedom 3 1 1 1 32 35. Sum of square 260238.04 5017.04 605 254616 2010.85 262248.89. Mean square 86746 5.017 605 254616 62.8 -. F value 1380.449 79.8396 9.6278 4051.880 <.0001 -. Pr > F <.0001 <.0001 0.0040 <.0001 -. Modelling of Thrust Forces in Drilling of AISI 316 Stainless Steel Using Artificial Neural Network and Multiple Regression Analysis. Contribution % 1.91 0.23 97.08 0.78 -. 495.
(737) Strojniški vestnik - Journal of Mechanical Engineering 58(2012)7-8, 492-498. Fig. 3. Comparison of experimental and residual thrust forces with thrust forces predicted by regression analysis. ;
(738) = 2;
(739) = ;
(740)
(741) ;
(742)
(743)
(744) 1
(745) H/H = ;
(746)
(747) ;
(748) = = !H!' !H3 ;
(749) !H
(750) 1
(751) H /H 6
(752)
(753)
(754) == = !H =
(755)
(756)
(757) = H
(758)
(759)
(760) . a). 6
(761)
(762)
(763) ;;
(764) 2;
(765) H
(766) &;
(767) =
(768)
(769) 2;
(770) H ;
(771) H<
(772) ;=
(773)
(774)
(775) ;
(776) =
(777) ; =[[H6
(778)
(779)
(780)
(781) =
(782) H6 /
(783)
(784)
(785) = =[[
(786) =
(787) H 6
(788) = =
(789)
(790)
(791) I H *0+H= =
(792) = %. b). 1 ⎛ Ff = ⎜ − ( 3.2694× F1 + 3.0517× F2 + 3.7990× F3 − 7.7062× F4 + 0.0511× F5 + 0.9165 ) ⎝ 1+ e ×950,. ⎞ ⎟× ⎠. *0+. Fi *i ´ 3 HHH / 4+
(793) I H*'+%. c) Fig. 4. Comparison of the actual and predicted thrust forces; a) 0.08 mm/rev, b) 0.1 mm/rev, c) 0.12 mm/rev. 496. {| }~` }
(794) }{. Fi =. 1 , 1 + e − Ei. *'+.
(795) Strojniški vestnik - Journal of Mechanical Engineering 58(2012)7-8, 492-498. Table 4. Statistical evaluation for the thrust force Goal. Learning algorithm. Number of neurons. Ff. SCG. 5. RMSE 0.00769. Training data R2 0.999907. Ei
(796)
(797) =
(798) ;
(799)
(800)
(801)
(802) 6 &H 6 =
(803) ;
(804)
(805) H 6
(806)
(807) ;
(808)
(809)
(810) 6 4H. MEP 0.725947. RMSE 0.009066. Testing data R2 0.999871. MEP 0.930127. = ;
(811) [[H1
(812) H4 ;
(813) = ;
(814) = [[H Table 5. The weights between the input and hidden layers i 1 2 3 4 5. Ei = w1x(Ct/6) + w2x(V/60) + w3x(fx4) + i w1 w2 w3 -6.7500 -4.4407 -6.8930 -2.3814 -6.5059 -4.7796 -9.1133 -4.9393 6.0572 0.0480 0.1832 -6.4262 3.3620 -8.5526 -2.2916. i 11.7137 11.8331 -2.7930 3.7340 8.9342. 3 CONCLUSIONS
(815)
(816) ;
(817) [[ ;; ;
(818) =
(819)
(820)
(821) = &
(822) H6
(823)
(824) <
(825)
(826) =
(827)
(828) H 6 ;
(829) =
(830) ;
(831)
(832) ;
(833)
(834) H ;
(835) = 2;
(836) H6;; ==
(837) ;
(838) =
(839)
(840)
(841) ; H6R2 = !H=
(842)
(843)
(844)
(845) [[
(846)
(847) !H3 =
(848) ;
(849)
(850) H ?
(851) = ;
(852)
(853) = =
(854)
(855) ;
(856)
(857) H
(858)
(859)
(860)
(861) ;
(862) ; =
(863)
(864) [[H6= = [[
(865)
(866) = ;
(867)
(868) ==
(869) = ;2
(870)
(871) 2;
(872)
(873) H. a). b). 4 REFERENCES c) Fig. 5. Comparison of the actual and predicted thrust forces; a) 0.08 mm/rev, b) 0.1 mm/rev, c) 0.12 mm/rev. =
(874)
(875)
(876) R2 =
(877)
(878) !H!0
(879) R2 =
(880) !H'0H69>I =
(881)
(882)
(883) ;
(884) !H!!0 !H!!! !H034/0 !H&!30;
(885) H6=.
(886) ; HH 9;
(887) H R ; H W H *3!!+H
(888) =
(889)
(890)
(891)
(892)
(893)
(894) H International Journal of Machine Tools & Manufacture H / H &/ ;H 343! Z%!H!#H
(895) # H3!!'HH!!H 3 1
(896) 8 H [#
(897) 1H 7
(898) 8 IH *3!!'+H I== = ;
(899)
(900) .
(901) = = >3 R
(902) H Journal of Materials Processing Technology, H 3! H & ;H /0/03 Z%!H!#H# ; H3!!0H3H!03H. Modelling of Thrust Forces in Drilling of AISI 316 Stainless Steel Using Artificial Neural Network and Multiple Regression Analysis. 497.
(903) Strojniški vestnik - Journal of Mechanical Engineering 58(2012)7-8, 492-498. & WH 6H$H *'+H Z;
(904) % HHeat Treating;H3'&3H /
(905) IHH 7
(906) HH *'!+H Cold treatment of cutting tools., All-Union ScientificResearch Tool Institute,H!;H40H*
(907) I
(908) + 4
(909)
(910) <H*+H<
(911) ;
(912) |
(913)
(914) H Electrical Discharge Machining Today, ;H&//H R WHH Q H6H "
(915) ¹HQH ?
(916) HWH ? >HH >
(917) 6HIH *3!!&+H >
(918) =
(919) >3 H Materials Science and Engineering A, H && [H 3 ;H 3/3// Z%!H!!34!&*!3+!!4¹H 0 Z
(920) 1HWH 1 ZHZH > H9H I8H9H 8 WHH>H *3!!+H $= =
(921) RHWear,H3;H0/ '4Z%!H!#H H3!!H!H!0H ' 6 H *+H <
(922) =
(923)
(924)
(925) =
(926)
(927) H Journal of Materials Processing Technology, H 3& ;H &//&/ Z%!H!!3/!&*+!!'&H R H Q ¹H "
(928) ¹H 6 H *3!!+H $
(929)
(930) =
(931) = = =;;
(932)
(933) =
(934)
(935) H Journal of Materials Processing Technology, H '! ;H 3&3 Z%!H!#H# ; H3!!H!/H3&H !Ó8 6H 7 ; H *3!!4+H $
(936)
(937)
(938) = =
(939)
(940)
(941)
(942) H International Journal of Machine Tools and Manufacture,H/4 H/4;H /0/0Z%!H!#H
(943) # H3!!/H!H!!0H $ HH < ZH $ H7H *3!!0+H >
(944)
(945) =
(946) =
(947) =88 ;;
(948) . International Journal Advanced Man ufacturing Technology,H&/ H&/ ;H3303&4Z%!H!!0!!0!!!!4'!H 3
(949) H >H *3!!+H >
(950) ;
(951)
(952) = =
(953)
(954) ;
(955)
(956)
(957) =
(958)
(959)
(960) ;
(961) . 498. HExpert Systems with Applications,H&' H 4;H4'34'&3H &[ >H » RH 6 ÔH H *3!!+H 6 2;
(962)
(963)
(964)
(965) = == = $Z <Z
(966)
(967)
(968) ; =
(969) <[<
(970)
(971) ;
(972)
(973)
(974)
(975) =
(976)
(977) H Robotics and ComputerIntegrated Manufacturing, H 34 H ;H 333& Z%!H!#H
(978) H3!!0HH!!/H /"
(979) HH " ?HH *3!!+H >
(980) =
(981) =
(982)
(983) H Journal of Materials Processing Technology,H!' H&;H 3'&Z%!H!!3/!&*!!+!!'&4H 47 ZH *3!!+H $
(984)
(985) = =
(986) <[<
(987)
(988) =
(989)
(990) . HJournal of Materials Processing Technology, H 3! H 0 ;H &34&&0 Z%!H!#H # ; H3!!'H!0H!3&H
(991)
(992) 6H H H 7 6H *3!!+H ;;
(993)
(994) =
(995) =
(996)
(997)
(998) ;
(999)
Benzer Belgeler
Vezin, redif ve kafiyenin ritim imkânlarını iyi kullanan Sâfî, şiirinde tekrarları da önemli bir ahenk unsuru olarak kullanmış, kelime veya kelime grubu seviyelerinde
O dönemin entelektüel bir kişiliği olan Defterdar Sarı Mehmed Paşa, devle- tin mevcut durumu üzerine kafa yorarak düzende oluşmuş bozukluklar için çözüm önerileri
Her ay muntazaman gönderdiğimiz mecmuamızın, 1941 yılı başlangıcından itibaren aylık ola rak intişar ettiğinden, iyi veya kötü bir muhtevaya malik
Bu dergiler; Çocuklara Talim dergisi, Çocuklara Mahsus Gazete dergisi, Yeni Yol dergisi osmanlıca olup; Atlas Çocuk dergisi, Bilge Çocuk eğitim dergisi ve Bilim Çocuk
The fatty acid (FA) composition of total nutlet lipids and tocochromanol contents of some Salvia species naturally growing in Turkey were deter- mined.. The results of the fatty
In the end, from both synthetic and real- life experiments we have conducted, we reached to the conclusion that gradient- equalized layers activated using bounded functions such
Çarpık bir ücret yapısı, işletme ve genel ekonomi düzeyinde insan kaynaklarının israf edilmesine ve toplumsal refah kaybına neden olmaktadır.89 Bu sebeple, rasyonel bir
Her fırsatta eğitime en çok payı kendilerinin ayırdığını iddia eden AKP hükümetleri döneminde eğitim bütçesinin milli gelire oranı OECD ortalaması olan