The cutting tool that is based on confluence of information of much parameter state wears away condition intelligence identifies

  • Time:
  • Click:145
  • source:ELAND CNC Machining
1 overview cutting tool wears away is the phenomenon that produces necessarily in metallic cutting process, spot statistic proves, cutting tool invalidation, damaged is to cause NC kind utilization rate of automation treatment machine tool drops, the chief factor that manufacturing cost rises, it is machine tool equipment and the main cause that cutting machines safe accident, stop by what its cause machine the 1/5 ~ 1/3 that takes breakdown machine down time. Because this begins cutting tool tatty online detect all in all real sense is had in automation treatment. Of cutting tool condition detect the method basically has direct way and indirect method. Direct way is to detect directly cutting tool wears away the numeric size of the face; Indirect method is wear away with cutting tool through detecting the physical signal related condition will reckon cutting tool wears away condition. All sorts of current monitoring technologies basically are to be centered at detecting secondhand. Among them most common relatively effective also cutting force detects, sound blasts off detect and electric machinery electric current / power detects etc. Cutting force is the mainest feature of token cutting process all the time, the change of cutting process is closely related to cutting force, with cutting force so condition of the cutting tool that monitor is having sensitivity tall, interference rejection sex is good wait for relatively outstanding advantage. The wear extent of cutting tool not only related to treatment operating mode, and the length that uses time with cutting tool is having close relationship. The article basically sets out from pragmatic point of view, built be based on cutting force wears away with two kinds when be based on opposite cutting time model. Cutting force model uses recursive algorithm and ambiguous classification technology to build, this model described parameter of cutting force, cutting and cutting tool to wear away the relation between condition, through detecting cutting force signal but online identifying cutting tool wears away condition. Be based on model of opposite cutting time to use recursive technology to establish the relationship of cutting tool wear extent and cutting parameter and time directly, can come true to identify to cutting tool tatty inside bigger cutting condition span. Two kinds 2 wear away identify what square legal principle considers a model to build when breakdown appearing in should machining a process (wear away like cutting tool or damaged) , the change that causes treatment status signal. Because this traditional supervisory system is of the diagnostic parameter from detected condition signal,breakdown feature is gotten in change, the identifying that realizes pair of treatment status breakdown then and decision-making. But, in machining a process the change as a result of cutting parameter also will cause the change that processes condition signal, the change that this kind of change often compares the treatment status signal that brings about by treatment breakdown should be gotten greatly much. Because traditional supervisory system fails change of condition of cutting of very good consideration is right the affect and can monitor onefold operating mode falls only treatment process that processes condition signal, and cannot use at advanced the treatment process with changeful condition of the cutting in making. At this point, the author uses faintness to return to network technology on one hand, built the mathematical model that is based on cutting force signal according to wear extent, use cutting force signal to judge cutting tool to wear away secondhand condition; On the other hand, use recursive technology, was based on opposite cutting time to build the mathematical model of cutting tool wear extent and cutting parameter and opposite cutting time. Graph 1 cutting tool wears away condition is subject spend function to wear away condition faintness classifies cutting tool to wear away is relatively slow process, normally wear process cent is 3 phase, namely it is level, normal that initial stage wears away wear away level and wear away quickly level. 3 phase are belonged to wear away differently condition, the bounds between each condition has certain ambiguous sex and jackknife sex. To automation treatment, detect cutting tool wear extent must be not measured accurately, it is certain to should know its are in only wear away inside limits can. Be aimed at milling, wear away state of affairs is very intricate, according to the requirement of milling process and experimental data, wear away cutting tool state branch is A, B, C, D, E5 kind, of all kinds average wear extent is respectively: 0.

1mm, 0.

15mm, 0.

2mm, 0.

25mm, 0.

3mm. The basis wears away the ambiguous sex of condition, if the graph is shown 1 times,build wear away condition is subject spend function. Choose B kind because add man-hour normally,be echelon is, in this interval its faintness degree asks relatively some lower. Be based on cutting force signal to wear away model cutting force is token cutting process the physical quantity of the mainest feature. The most change that produces in cutting process is mixed closely related cutting force. Study a proof in great quantities, increase as cutting tool tatty will bring about cutting force to increase, cutting tool damaged changes those who cause cutting force amplitude suddenly. The change of Ap of deepness of F of rate of cutting speed V, feed, cutting is met the size that affects cutting force. In the meantime, what the size of cutting force also follows the specific cutting environment such as workpiece material, cutting tool material is different and differ somewhat. According to metallic cutting principle, to Xin Ren the relation of force of cutting tool cutting and cutting dosage is like next formula: In F=Cvxfyapz (1) type: C -- the coefficient that decides Yu Dao has geometrical dimension and material quality; X, y, z -- the index of cutting dosage. Type (the concern that 1) expressed to force of the cutting in machining a process is the same as each cutting parameter, model of this cutting force belongs to a static state to be not line form model, after longitudinal form is changed, the model turns into: InF=InC+xInv+yInf+xInap (2) is aimed at different wear away condition cutting tool, in A, B, C, D, E5 kind in, be like next expression accordingly type: A kind: Sa=InFa=a11+a12Inv+a13Inf+a14Inap B kind: Sb=InFb=a21+a22Inv+a23Inf+a24Inap C kind: Sc=InFc=a31+a32Inv+a33Inf+a34Inap D kind: Sd=InFd=a41+a42Inv+a43Inf+a44Inap E kind: Se=InFe=a51+a52Inv+a53Inf+a54Inap(3) uses matrix form expression to be: S=W*X (4) pursues signal of network of model of 2 cutting force is schematic (4) can use a figure the nerve network form of 2 will convey. The BP algorithm that does not use a convention to this network model will adjust weight value, use recursive technology to adjust directly however weight value, because the relation between output of this network input has been linear,this is, and the study rate of the nerve network that is based on regression is rapid, OK and online study identifies. Cutting force model is the eigenvalue that is a model with cutting force amplitude, below the circumstance of foregone V, F, Ap, use formula (3) is pushed so that Fa, Fb, Fc, Fd, Fe regards a graph as 1 medium faintness get together kind center, the cutting power worth that will detect and get together kind of center quite, will decide current cutting tool wears away thereby subject spend and achieve detect purpose. Those who be based on opposite cutting time wear away the model wears away according to machining the cutting tool in the process transcendental knowledge of the rule is knowable, cutting tool wear extent follows the development of handling time and change. Below different cutting condition, the speed of growth is different, consider to discover, below same cutting condition, the metabolic ratio of wear extent is fair value approximately, but be based on operating condition consideration, the relation of the change of the change of wear extent and time fixed position is exponential relation. Establish formula according to formula of cutting tool life again (5) place shows those who be based on opposite cutting time wear away detect model. The input quantity of the model is cutting speed V, feed speed F, cutting deepness Ap, dt of initiative time-interval of cutting of VB0 of cutting tool wear extent, model output is cutting tool wear extent appraise measures value VB(t) . DVB(t)=VB(t)-VB0=KwvxfyapzDtm (5) linearization is: In InDVB(t)=InKw+xInv+yInf+zInap+mInDt (6) type, dt, wear away below same cutting condition detected cutting time-interval; DVB(t) , same cutting condition issues certain time-interval inside -- the change value of wear extent; Kw, the coefficient that concerns with dimension of cutting tool geometry and material property. Graph system of test of monitoring of 3 milling cutting tool 3 wear away model coefficient establish and device of experiment of experimental test and verify and experimental condition test have on HURCO-BMC-20 vertical machining center. A Kistler9257A 3 to the appearance that measure power, amplifier of 3 YE5850 charge, enlarge multiple is the tool microscope of 40 times, personal computer of a PC586. Experimental principle sketch map is shown 3 times like the graph. The signal classics of sensor magnifies and sampling and processing will be undertaken by personal computer control after the low wave that connect filter, because milling process has tooth cutting and cutting tool turn two cycle, the method that implements as synchronous as cutting tool turning period sampling to signal will obtain every to turn the size of average milling force. It is research target with milling cutter of keyway of blade of two tine helix, undertake checking an experiment with different cutting parameter, identify model coefficient according to actual measurement data. If experimental condition expresses 1 to show. Express high-speed steel of F10 of milling cutter of keyway of fluid of cutting of Mm of deepness of cutting of Mm/min of speed of feed of M/min of speed of cutting of material of workpiece of means of milling of material of cutting tool of condition of 1 milling experiment to arrange mill stainless steel 21.

98 35 3.

5 without 18.

84 30 3 15.

70 25 2.

5 models coefficient is gotten the rationality that reachs its desired result to establish a model for place of test and verify and practical, had many test. Graph 4 it is to use wear extent to be respectively 0.

15, 0.

2, 0.

25, 0.

The cutting tool of 3mm, the training of milling force model that undertakes below the cutting condition of different combination laboratory gets uses sample, in all 45 groups. Every curve in the graph undertakes below the cutting condition of same combination. Use recursive algorithm to get network advantageous position to be worth matrix to be: The correlation coefficient R2 that afore-mentioned weight are worth every to go is respectively 0.

9871, 0.

9765, 0.

9654, 0.

9453, 0.

9645. Because this can see a milling strength that found,the model mirrors the affinity that gives force signal and cutting parameter. Pursue training of model of 4 cutting force pursues with example training of 5 time model pursues with example 5 it is to be below different cutting condition, the sample book of 36 groups of training that the cutting tool that initial value wears away to differring inside different time-interval has laboratory is obtained. Every curve in the graph also undertakes below the cutting condition of same combination, use recursive algorithm training to get form directly, 6, population is ordinal for - 2.

1348, 0.

8607, 0.

1111, 0.

6143, 0.

8333, correlation coefficient R2=0.

9857 make clear wear away metabolic quantity and cutting parameter and opposite cutting time have closer relationship. To undertake test and verify to afore-mentioned models, watch 2 listed concrete desired result uses data sample. Desired result result is like a watch respectively 4 reach a figure 6, the graph is shown 7 times. Watch 3 with the graph 6 watches understand milling force model to wear away to cutting tool the category has taller identifying accuracy rate. Graph 7 it is normal to show cutting tool is in wear away when level, the model has taller identifying to lead; Wear away in initial stage and wear away quickly level, wear away distinguish effect difference. But because effective treatment process is medium, we care wear away condition basically is to be in wear away normally level, so this model can apply on the project. But of the not stable element as a result of effective treatment appear (wait for) like hard particle, time model cannot give reflect, when applying actually so, answer to use cutting force model and time model band. Express 2 desired result to use initiative wear extent of N of force of cutting of actual measurement of Min of time of cutting of Mm of deepness of cutting of Mm/min of speed of feed of M/min of speed of cutting of example serial number Mm121 of Mm real wear extent.



























































The wear extent of 235 actual measurement of serial number of desired result result that express model of 3 cutting force Mm actual measurement wears away the category is subject spend appraise to measure wear away category appraise measures wear extent MmµAµBµCµDµE10.


































234 categories identify rate 84.

6% graphs graph of effect of desired result of model of 6 cutting force pursues effect of desired result of 7 time model pursues 4 conclusion the article basically sets out from pragmatic point of view, built be based on cutting force wears away with two kinds when be based on opposite cutting time model. Main conclusion is as follows: Use the milling force model that ambiguous classification and recursive analysis place build, bona fide reflected the concern of cutting parameter and cutting force, wear away relatively ideally to cutting tool condition undertakes classify and discriminating. Desired result result makes clear, the model is right of cutting tool condition identify effectively rate for 84.

6% , there is applied value on the project. Discussed those who be based on time to wear away the method that build a model. Use recursive analysis, establish the relationship between cutting tool wear extent and cutting parameter and time parameter directly, desired result result makes clear, this model can satisfy the requirement on the project on certain level. CNC Milling