Summary: The article uses harmony of law of machine tool power to launch a law to undertake monitoring to the cutting tool damaged in turning process. The form of 4 kinds of expression that discovered signal of power of bed of opportunity of cutting tool damaged in the experiment, showed the random sex of form of cutting tool damaged. Be aimed at this kind of circumstance, offerred the way of delay time variance that power signal handles first; Blast off to all sorts of sound that give out in cutting process (the diagnostic amount that analysis of the frequency when AE) signal is used undertakes handling and collects damaged of reflective cutting tool, what used nerve network ART2 to achieve status of cutting tool damaged finally is automatic identify. Monitoring of cutting tool condition is one of crucial technologies that assure product quality and treatment system to move reliably. The method of damaged of the cutting tool that monitor basically has harmony of law of cutting force law, power of vibratory drilling method, machine tool to launch a law to wait. Because cutting force law measures power plant to need to change machine tool structure constantly in installation, accept not easily for user place. Compare with photograph of cutting force law, law of machine tool power has signal to measure handy, can avert the interference that bits, oily, vibration cuts to wait in cutting environment, the advantage that monitoring and warning device installs easily. Consider to make clear: When cutting tool damaged, extruding is mixed in cutting tool between workpiece rupture the accretion that the fragment can cause cutting strength, next cutting force is decreased along with fractional desquamation again for 0, contact workpiece again when cutting tool when, cutting force adds suddenly again. But the author discovers in experiment process, when cutting tool damaged, the electromotor power that reflects secondhand as cutting force has more metabolic forms, mirrorred the random sex of cutting tool damaged. Sound blasts off (AE) signal originates directly cutting area, have frequency tall, acute, answer fast characteristic, special agree with the monitoring of cutting tool damaged. But the acoustical emission source in cutting process is very much, damaged of the burst that is like material, cutting tool, cut bits to break off, those who cut bits and work bump etc, because how this is opposite,transmit undertakes sound be handled effectively, from numerous sound the extraction in transmit gives sensitive to cutting tool damaged feature to measure is a crucial question. In view of above reason, the article used harmony of law of machine tool power integratedly to launch Falaishi to show the monitoring of cutting tool damaged. The delay time variance that the power signal that the author puts forward handles introduced above all in article, the frequency when introducing what AE signal handles at the same time analyses a way, built system of test of cutting tool damaged, the signal of cutting tool damaged that uses afore-mentioned methods to arrive to collecting undertakes handling, extraction gives the diagnostic volume that reflects cutting tool damaged, those who used nerve network ART2 to realize cutting tool damaged is automatic identify. When 1 signal handles damaged of cutting tool of law of delay time variance, can cause signal of machine tool power to produce change. Because change of amplitude of power signal time domain has uncertainty, because this author formulated a kind of new data processing method, say for law of delay time variance, will collect the diagnostic amount of cutting tool damaged. Cutting tool is when normal cutting, inside a paragraph of very short time, the condition of cutting tool can think to keep changeless, signal is smooth. When computational signal variance, calculate above all of signal of the power when normal cutting all be worth, calculate after the delay time via period of time next the variance of power signal, calculate namely the signal that signal variance place uses all is worth is the power signal before period of time all is worth. The advantage of such processing is, after cutting tool damaged, if use the signal of current hour to all be worth, criterion the variance that variance is the signal after cutting tool damaged, although variance of the signal after cutting tool damaged has increscent tendency, but opposite for increscent rate is not very remarkable, and the variance that signal of the power when using damaged all is worth to signal of the power when normal cutting, can more the change that reflects a cutting tool condition explicitly. Above is the basic idea of law of delay time variance, specific algorithm is as follows: The signal length N that decides computational signal all is worth E and signal variance D above all and L, reach the D of delay time length that all is worth difference of the other side. The power signal that sets T hour to collect is X(t) , when T=t+1 hour, with the following recurrence formula computation new signal all is worth E and variance D: E(t+1)=E(t)+1x(t+1-d)-x(t+1-d-n)]n(1)D(t+1)=D(t)+1[(x(t+1)-x(t+1-l)] × [X(t+1)+x(t+1-l)-2E(t+1)n(2) is like D to increase substantially, criterion cutting tool produced damaged. The AE signal that frequency analyses the place in law cutting process to give out when can be divided for successive model and dash forward hairstyle two kinds, dash forward the amplitude of hairstyle AE signal often more successive model big multiple count even decuple. Generation is broken out model the acoustical emission source of signal is more, those who include micro-crack of interior of cutting tool damaged, cutting tool is patulous, cut bits to break off, cut bits and work between bump etc, because this is mere,from time domain angle the signal distinction that cannot give out the AE signal that issues by cutting tool damaged and other AE source leaves come. And the essential hypothesis that analyses a method for fundamental frequency domain with FFT is AE signal it is smooth or it is changeless signal, break out however model AE signal often expands with material interior crackle, material ruptures etc closely related, it is one kind is not smooth signal, because this more reasonable method is,from time domain and frequency domain two respects analyse the metabolic case of AE signal at the same time. This literary grace distributings with the index that offers by Cloi-Williams ED(ExponentialDistribution) . Cohen gives out when the unified expression form that frequency distributings is as follows: Cx(t, w, ~)=1ej(xm-tw-xt)Ø(x, in type of T)x(µ+t) × X*(µ-t)dµdtdx2p22(3) : X(µ) is time domain signal; X*(µ) is its answer conjugate; Ø(x, t) is nuclear function (Kernal Function) , give different &216;(x, t) can get differring when frequency distributings. Cohen kind distributinging is bilinear distributings, using this kind to distributing when the frequency when undertaking is analysed, produce across inevitably interference problem. And the exponential nucleus Ø(x that puts forward by Cloi-Williams, t)=e-x2t2/s(4) pursues the composition of 1 test system (A)(b)(c)(d) pursues all sorts of change forms of signal of the power when damaged of 2 cutting tool solved this one problem well. ED is right across restraining is to pass the volume that adjusts invariable S to come true. The S=1 that uses when the frequency when undertaking to AE signal is analysed. If the graph is shown 1 times,system of test of damaged of 2 experiments cutting tool is comprised. Use J1-MAZAK530 center lathe, 45 steel bar expect and but razor blade of dislocation hard alloy (YT15) . If plan institute is shown, power changes send implement send A/D sampling board changeover of electromotor instantaneous power input into voltage signal, use frequency to be 4kHz, sampling length is 8000 a little bit. Sensor of pottery and porcelain of AE pressing report is originallied closely to approach knife head office on arbor, with the AE signal that happens in collect cutting process. AE signal enters oscillograph of THS720 high speed via evacuation of enlarge filter wave, signal sampling frequency is 2. 5MHz, sampling length is 2500 a little bit. Because be below normal cutting condition, damaged of cutting tool nature happens very hard, to quicken the damaged of cutting tool, use the damaged that infiltrates in workpiece the method of hard particle will quicken cutting tool, hard particle is chosen for head of high-speed steel bradawl. To get area of different cutting tool damaged, take different rotate speed respectively, cut mix greatly feed has a test, collected the power signal when damaged of many cutting tool and AE signal. 3 data analysis and result test result make clear, the change of signal of the power when cutting tool damaged has certain random sex, in the experiment we discovered 4 kinds of circumstances (like the graph 2 show) : (After damaged of A) cutting tool, power signal increases quickly, this may be caused; (B) power signal rises first the circumstance that immediately drops again, because a place of strategic importance of cutting tool fragment of the damaged when cutting tool damaged is squeezed between cutting tool and workpiece to cause power to rise,this may be, but immediately falls off again and cause power to drop; (The circumstance that C) power signal drops quickly, this has two kinds of possible reasons, it is the damaged with cutting tool bigger happening, make cutting deepness has bigger reduce cause power to reduce thereby, 2 it is the knife before cutting tool happening face or hind knife face flakes, or the horn after around knife face flakes the horn before causing effective cutting is mixed at the same time increases, cause thereby use up power to reduce; (The circumstance that after D) power signal drops first, rises, cutting tool is in the instant that this may be cutting tool damaged to cut position for nothing, make power signal shows the phenomenon that increases again after the instant drops. Among them (the case that A) place shows produces probability taller. Graph 3 it is corresponding at the graph 4 kinds of 2 circumstances issue the result that uses processing of law of delay time variance, can see from inside the graph, when damaged of cutting tool happening, variance of signal delay time increases substantially. The damaged that this kind of means that because this uses the author,offers will come to to monitor cutting tool is judged easily. The delay time variance of power signal can regard the feature of cutting tool damaged as the quantity. (The signal of 4 kinds of typical AE that A)(b)(c)(d) graph detects in the process of experiment of delay time variance of 4 kinds of signal when damaged of 3 cutting tool, if the graph is shown 4 times, its are corresponding when frequency distributinging graph is shown 5 times like the graph. What from OK and apparent ground sees 4 kinds of signal in the graph frequency distributings is different. Frequency distributinged to offer more at judging the information of cutting tool condition using than time domain or frequency domain analysis when, go up in frequency domain and time domain cannot accurate 4 kinds of signal that undertake classified, in when frequency domain is having good classified feature. To undertake discriminating automatically, distributing to draw a feature by the following measure for the basis with the frequency when AE signal: Go to time and frequency regularization inside unified time-domain T and frequency region F. Break up time domain equidistance for M stature region, limits of every stature region is F=F/n, break up frequency domain equidistance for N stature region, limits of every stature region is F=F/n, computation the energy Ui that frequency domain of every period of the day from 11 pm to 1 am is the place inside T × F to have, j, i=1, 2, ... , m; J=1, 2, ... , n. By Ui, j forms U of vector of dimension of M × N, undertake vector U normalization handles: U={Ui, j=Ui, j/max(Ui, j) , i=1, 2, ... , m; J=1, 2, ... , the AE signal when N}(5)(a) normal cutting (the AE signal when damaged of B) cutting tool (the AE signal when C) cutting ruptures (the AE signal picture that D) produces randomly the AE signal of 4 different types (A) is normal the AE signal when cutting (damaged of B) cutting tool, j. Although the diagnostic amount that receives with this kind of method is having the circumstance of overlay to fall to also can identify the damaged of cutting tool correctly in signal, the damaged that because of the frequency when this uses what the author puts forward the analysis identifies cutting tool is very effective and reliable. Of the quantity of feature of delay time variance power signal and AE signal when composition of quantity of frequency distributinging feature combines vector, the input comes to get used to resonance network ART2 oneself, can identify the damaged that gives cutting tool automatically, identify successful rate for 97.
5% . The article used 4 brief summary the monitoring that harmony of law of machine tool power launchs Falaishi to show cutting tool damaged, built system of test of cutting tool damaged. The form of 4 kinds of expression that discovered signal of power of bed of opportunity of cutting tool damaged in the experiment, showed the random sex of form of cutting tool damaged. Be aimed at this kind of circumstance, put forward law of delay time variance to process power signal, data analysis shows this kind of method is feasible as a result. Be aimed at the AE signal that issues in cutting process, the method that the frequency when using analyses undertakes handling, extraction gave the diagnostic amount of damaged of reflective cutting tool, what used ART2 to realize cutting tool damaged is automatic identify, identify successful rate to achieve 97.
5% . CNC Milling
CNC Machining