Foreword in recent years, the trend as product miniaturization and microelectronics machine system (MEMS) flying development, grow day and day to the demand of small small-sized structural member. Imperceptible cutting technology is machining respect of small small-sized and three-dimensional structural member to have treatment precision tall, efficiency tall, cost is low, three-dimensional process capability is strong, applicable workpiece material range is wide wait for an advantage, among them imperceptible milling technology is practical the strongest, application is the widest, become the key that machines research and heat. If each function that the exterior surface roughness after finish machining is worth pair of spare partses works stiffness of precision, contact, cooperate the influence such as function, corrosion resistance, wearability very big, especially to small small-sized structural member, size of its oneself geometry is small, accurate ground is forecasted and the exterior surface roughness after treatment of control spare parts has crucial real significance and economic value. With groovy measure machinery cutting photograph is compared, imperceptible cutting companion has appearance of ply of the smallest cutting, its are not free cutting degree to also be strengthened accordingly, appear thereby scale effect phenomenon, make the basic rule of imperceptible milling is different from groovy milling. Study imperceptible milling condition falls, the craft parameter influence to exterior surface roughness the rule, and the cutting parameter before accordingly logical choice is machined, make control and be forecastinged appearing be weighed particularly to exterior surface roughness should. The article uses orthogonality to return to a method to build exterior surface roughness to forecast a model above all, the parameter of relevant cutting craft that decides with this model place next is a foundation, use 2 times answer compose of curved surface law to build optimize forecast a model, undertake relevant experiment studies. 1 orthogonality returns to those who forecast a model to build imperceptible milling to machine a condition to fall, feed of milling rate V, every tine Fz, the α P influence to exterior surface roughness has cutting deepness to differ apparently, use orthogonality to return to a method to build the exterior surface roughness between exterior surface roughness and milling parameter to forecast a model to be the C in type---Milling condition amends coefficient B1, B2, B3 with what machine material decision---The index is in type (1) both sides takes natural logarithm to get LnRa of your Y ＝ respectively, lnv of X1 ＝ , lnfz of X2 ＝ , p of α of Ln of X3 ＝ , lnC of B0 ＝ , criterion type (2) can adapt to use matrix form to express to be ε of ＋ of Y ＝ XB to build equation of multivariate and linear regression (4) use a least square method to undertake estimating to parameter B, can beg a population B0, B1, B2, B3, can beg exterior surface roughness to forecast a model instead then examine related hand-in-hand travel. Method of 2 RSM model answers curved surface way (Responsesurfacemethodology, abbreviation RSM) can discuss factor of influence and answer the mathematical concern between output. One is included answer Y (X) process or system, should answer rely on input factor X1, x2, ... , xp. Those who consider alternant effect and 2 effect 2 rank I of the β in answering curved surface model to be able to express to be type---The slope of encode Xi or Ij of linear effect β---The Ii of alternant effect β between Xi and Xj---The β of 2 effect of Xi 0---Invariable can write into 3 2 to plant to answering curved surface model 2 times of 3 elements machine tool of the disparate test design of the model and test result experiment is machining center of small small-sized Che Xi, drive system repeats fixed position precision to be 0 2 μ M, resolution is 0 1 μ M, top rotate speed is milling report main shaft 60000r ／ Min, workpiece material is duralumin alloy LY12, cutting tool uses a diameter to establish milling cutter for the hard alloy of 3 blade fine grain of 1mm. The measurement of exterior surface roughness uses exterior surface roughness appearance (times TR240) undertake. Deepness of imperceptible milling cutting is in commonly 10 - 4 ～ 10 - inside 2 Mm limits. To eliminate an error, the cutting that must have 0 section first is generated, decide according to the 0 tangent plane that already formed the experiment uses cutting deepness next, undertake planar milling experiments next, in order to assure the accuracy of the result. Orthogonal regression forecasts a model in, experiment of imperceptible milling craft introduced 3 elements the orthogonal test program of 3 levels, if test data expresses 1 to show. Take logarithm to test result, of attainable surface surface roughness forecast a model to be based on orthogonality to rotate to answered curved surface model 2 times to use the program that sets a design, this design method needs to determine the 0 levels of the experiment above all. According to orthogonality regression forecasts the test program of the model, decide 0 levels fall ／ Min of 12000r of ＝ of N of main shaft rotate speed, feed is F ＝ Min of ／ of 4 5mm, cutting deepness is α P ＝ 30 μ M, 1mm of ＝ of D of cutting tool diameter, mill tooth counts Z ＝ 3, according to above parameter can have the consideration of milling rate and every tine feed. Variable to 3 the list can be gotten after deepness of feed of milling rate, every tine, cutting has code respectively namely 2 show an experiment to design standard matrix, have a test according to this, gotten test result. The suitable scope that answers curved surface equation 2 times when draft the least square estimation that close to forecast a model for this surface surface roughness is: 53 of ≤ of V of ≤ of Min of ／ of 21 84m Min of 52 M ／ ; 0 of ≤ of Fz of ≤ of Z of ／ of M of μ of 0 11 Z of ／ of M of 39 μ ; 46 μ of ≤ of P of α of ≤ of 13 μ M M. 4 2 is planted the comparison that forecasts a model examines reach test result to analyse pair of above of 2 kinds of methods forecast a model to undertake if the watch is shown 3 times,significance examines. The result makes clear: Answer curved surface law 2 times to forecast a model to show height remarkable, of law of regression of apparent excel orthogonality forecast a model, the exterior surface roughness that answers model of curved surface law to forecast LY12 of alloy of imperceptible milling duralumin with 2 has higher reliability. The result that forecasts 2 kinds model place and real test result undertake comparative, if pursue,1 is shown. From the graph 1 in can see, answer what curved surface forecasts a model 2 times to forecast a result to forecast a model to be close to the exterior surface roughness that gets at real test to be worth more than orthogonal regression, farther also confirm the result that significance examines. The returns to equation population that gains to answering place of curved surface method 2 times undertakes significance examines, can inspect every each variable to because of,be opposite the remarkable extent of variable action, be opposite thereby the milling craft parameter that affects surface roughness of imperceptible milling surface undertakes the result undertakes forecast and be controllinged analytic experimenting with be opposite better. Examine to show the significance sort that affects in surface roughness of the surface below condition of experiment of existing and imperceptible milling is as a result: ① linear effect: Milling rate (V) , cutting deepness (α P) , every tine feed (Fz) . ② 2 effect: 2p of F 2z, V2, α . ③ alternant effect: Vfz P of α of alternant, V Fz α P is alternant, alternant. According to type (8) model of exterior surface roughness undertakes scale pursues planning to incorporate 2 ～ 4, get each craft parameter is opposite the influence rule of exterior surface roughness. (1) graph 0 of the 2 ＝ that it is Fz Z of ／ of 25 μ M, namely every tine feed falls for condition of 0 levels milling, the response of milling rate and cutting deepness to exterior surface roughness curved surface. Cutting deepness 46 μ of 13 ～ inside M limits, exterior surface roughness is worth the addition along with milling rate to drop apparent. 60m ／ Min is close to in milling rate, cut deep most hour, exterior surface roughness is adjacent 0.
2 μ M. Raise milling rate to be affected to reducing exterior surface roughness it is thus clear that remarkable. (2) graph 30 μ of ＝ of the 3 P that it is α M, cut namely deeply below condition of 0 levels cutting, the response of milling rate and every tine feed to exterior surface roughness curved surface. In ＝ of V of low milling rate 18.
84m ／ Min, along with the addition of every tine feed, exterior surface roughness is worth change not remarkable. Dan Zaixi is cut when rate is older, exterior surface roughness is affected along with the addition of every tine feed remarkable. In milling rate when Min of ／ of 40 ～ 50m, exterior surface roughness is worth stability to be in 0.
Inside 4 μ M, the effect that suffers other parameter is lesser, right now, can acquire stabler treatment surface quality. (3) graph 4 for V ＝ when Min of ／ of 37 68m, namely milling rate falls for condition of 0 levels milling, the response of cutting deepness and every tine feed to exterior surface roughness curved surface. Curve degree to be able to see from what answer curved surface 2 times, the addition of cutting deepness and every tine feed makes the value increases exterior surface roughness significantly, among them cutting deepness ages than every to the influence of exterior surface roughness feed wants many remarkable, cutting be worth the least state of affairs greatly to fall with every tine feed, can obtain the smallest surface surface roughness to be worth. 5 conclusion (1) be based on orthogonality to return to what the law answered curved surface law to build surface roughness of the surface below imperceptible milling condition respectively with 2 to forecast a model, undertook relevant experiment studies, via examining 2 times to answer curved surface model to have inside experiment condition taller degree of confidence is mixed practical, forecast precision and method of regression of orthogonality of as apparent as the registration of actual result excel, comfortable be forecasted at cutting parameter choosing to undertake optimizing to exterior surface roughness before treatment and control as a result. (2) the significance of 2 the linear effect that analysed milling parameter to affect exterior surface roughness through answering curved surface 2 times to forecast model and test result, effect and alternant effect. Compare integratedly, imperceptible milling is affected to exterior surface roughness in 3 cutting parameter bigger is milling rate, it is every tine feed next, cutting deepness is the weakest to the influence of exterior surface roughness. Below experiment condition, raise milling rate, reduce every tine feed and cutting deepness to be able to reduce surface roughness of imperceptible milling surface effectively to be worth. (3) the exterior surface roughness that is based on orthogonality to rotate 2 times when combine a design to answer curved surface law to build forecasts a model, in have taller degree of confidence while, can add the information content of the experiment effectively, rise forecast precision, reduce experiment number, this kind of method also can be used at building to forecast a model otherly, but must notice its suitable scope. CNC Milling