Production is actual in, the factor that affects machining error often is daedal, analyse its very hard because of always with sheet sometimes causal, and should use method of several manage statistic to undertake integration analysis will find out the way that solves a problem. One, the property of machining error the machining error of all sorts of only factors, those who press its statistic rule is different, can divide for systematization sum of errors random sex error two kinds big. Systematization error is divided again be worth systematic error to often be worth systematic sum of errors to change two kinds. (one) L of of systematization error . After often be worth in order of systematic error to process a batch of work, its size and direction maintain changeless error, call error of constant value system. The production error that processes machine tool of principle sum of errors, clamping apparatus, cutting tool for example, it is error of constant value system. In addition, of machine tool, clamping apparatus and measure wear away rate is slower, also can be being regarded inside proper time is error of constant value system. 2. Change in be worth in order of systematic error to process a batch of work, its size and direction press the error that definite pattern changes, call change value system error. For example the machine tool, clamping apparatus and cutting tool heat before hot balance is out of shape of cutting tool of sum of errors wear away etc, it is to change value system error. (2) in order of of random sex error processes a batch of work, occurrence size and way are different and the machining error that changes without the rule, call random sex error. For example semifinished product error (surplus size differ, hardness is inhomogenous etc) answer mirror, fixed position error (differ, clearance affects datum plane precision) , clamp error (clamping force size is differ) , the error that adjusts for many times, leftover stress causes be out of shape the error, it is random sex error. Random sex error looks from the surface do not have what rule it seems that, but the overall pattern that the method that uses several manage count can find out machining error of a batch of workpiece, adopt measure to try to control on craft next. 2, the statistic of machining error analyses a way (distributinging graph analyses a way) statistic analysis is the data data that the spot observes and has in order to produce examining actually to workpiece is a foundation, the methodological analysis of statistic of the manage that use number handles these data materials, announce thereby all sorts of elements are right the combined influence of machining error, obtain the means of a kind of analysis of the way that solves a problem, basically distributinging graph analyses law and law of the analysis that choose a plan to wait. This section basically introduces distributinging plan law. Other method asks referenced concerned data. 1. Distributing actually graph -- histogram is in treatment process, to the treatment of some working procedure dimension uses data of draw-out and finite example to undertake analytic handling, the form that uses histogram expresses, in order to facilitate the analysis machines quality and the method that its stabilize rate, call histogram to analyse a way. In draw-out finite example data, the change that machines dimension calls dimension to disperse; The spare parts amount that appears in same size interval calls frequency; Of frequency and this batches of sample sum than calling frequency; Frequency and group are apart from (dimension is removed) than calling frequency density. The dimension with workpiece (a paragraph of very small size is removed) for the abscissa, what state for ordinate with frequency or frequency this working procedure machines dimension is actual distributinging plan says histogram, if pursue,4-32 place is shown. The group of rectangular × of density of area ＝ frequency on histogram is apart from (dimension is removed) because ＝ frequency is all the sum of each groups of frequency is equal to 100 ％ , the sum of all and rectangular area is equal to L on reason histogram. The course of action of histogram shows below through example: For example diameter of a batch of axes is grinding the workpiece of Mm, if the dimension after actual measurement expresses 4-3 place to show. Biao 4-3 Zhou Jing Chi Cun Shi Ce Zhi （μ m）44 20 46 32 20 40 52 33 40 25 43 38 40 41 30 36 49 51 38 34 22 46 38 30 42 38 27 49 45 45 38 32 45 48 28 36 52 32 42 38 40 42 38 52 38 36 37 43 28 45 36 50 46 38 30 40 44 34 42 47 22 28 34 30 36 32 35 22 40 35 36 42 46 42 50 40 36 20 16 Sm 53 32 46 20 28 46 28 54 La 18 32 33 26 46 47 36 38 30 49 18 38 38 Zhu : The data in the watch is the difference of actual measurement dimension and basic measure. The move that makes histogram is as follows: (1) collect data. Take 100 or so commonly, find out maximum La ＝ 54 μ M, 16 μ M(sees the smallest ＝ that be worth Sm watch 4-1) . (2) distribute 100 example data into a certain number of groups, count usable watch 4-4 to decide in group. The chooses data amount that expresses 4-4 example and set number counts 100 6 of 50 ～ in group this exemple of of 20 of ～ of 10 of above of ～ 12 250 of 250 7 of ～ of ～ 10 100 takes ＝ of set number K 8. Experience proves, set number can cover the fluctuant situation of inter block data too less, set number is too much can make each groups height uneven, do not look to give metabolic rule thereby. Normally certain set number wants every groups to make spread out on average at least 4 ～ 5 data. (3) computational group is apart from H, namely the removed of group and across block H = = =4.
5 μ of ≈ of M of 75 μ of M (4) calculate the first group on, next bounds are worth of Sm ± the first group upper bound is restricted to be worth for = of S M ＋ (16+ ) μ M = 18.
5 μ M; Next end worths are Sm - = (16 - ) μ M = 13.
5 μ M. (5) computational the others each groups go up, next end worths. The first group upper bound is restricted to be worth is next the 2nd group end worths. Next the 2nd group end worths are added group be apart from even if the 2nd group upper bound is restricted to be worth, the others analogize. (6) the center that counts each groups is worth X I. Central value is every groups of numerical value among. Xi = (some group of upper limit are worth + some group of floor level are worth) / the first group of centers are worth 2 of of M of μ of M=16 of Xi= Xi= μ (7) the data that records each groups, arrange Cheng Pin to count distributinging watch, if express 4-5 place,show. (8) each groups size frequency, frequency mixes statistic frequency density, in filling a list. (9) it is ordinate with frequency density by the data that express a kind; Group be apart from (dimension is removed) can draw a histogram for the abscissa, if graph 4-32 place shows a list,4-5 frequency distributings class boundary of N of watch set number / frequency of Mi of frequency of statistic of frequency of I of χ of value of μ M center / % frequency density / (μ M-1) (% ) 113.
5~18.
516 | | | 330.
6218.
5~23.
521 | | | | | | | 771.
4323.
5~28.
526 | | | | | | | | 881.
6428.
5~33.
531 | | | | | | | | | | | | | 13132.
6533.
5~38.
536 | | | | | | | | | | | | | | | | | | | | | | | | | | 26265.
2638.
5~43.
541 | | | | | | | | | | | | | | | | | | | 16163.
2743.
5~48.
546 | | | | | | | | | | | | | | | | | | | 16163.
2848.
5~53.
551 | | | | | | | | | | | | 10102953.
5~58.
556 | 110.
2 by graph 4-32 knowable, dispersive limits resides this measure that approves work for the most part in, slant big, slant small person less. Dimension is dispersive limits = is the biggest diametical – is the smallest diametical = 60.
054 – 60.
016 = 0.
Dimension of 038mm is dispersive limits center: The public errand of diameter of Mm takes central = 60+ = 60.
Standard deviation of of 025 Mm is: of = = Mm can see from inside the graph, the dispersive limits of this batch of workpiece is 0.
038, smaller than tolerancepublic errand belt, but dispersive limits center and public errand take measure center not coincide, if try to adjust dispersive limits center,take coincide with tolerancepublic errand, should increase the radial feed of the machine tool only namely 0.
012 Mm, can eliminate error of constant value system. CNC Milling