CN106217130A - Milling cutter state on_line monitoring and method for early warning during complex surface machining - Google Patents

Abstract

The invention provides milling cutter state on_line monitoring and method for early warning during a kind of complex surface machining, belong to field of diagnosis about equipment fault.The present invention demonstrate Milling Force and handle of a knife vibration current vortex displacement signal between there is linear relationship on the basis of, using machine tool chief axis end X that three-dimensional acceleration transducer records, the acceleration signal of Y-direction secondary frequency volume integration as handle of a knife vibration displacement signal, extract the fundamental frequency in integral displacement signal and harmonic signal thereof as monitoring signal, on the one hand overcome all drawbacks directly being carried out displacement detecting in Machining of Curved Surface by eddy current displacement sensor, on the other hand can effectively remove the impact of interference signal；Use Synchronous time average cumulant that monitoring signal is further processed, then utilize characteristic quantity quantitatively to portray synchronized averaging cumulant waveform, realize the monitoring to milling cutter state according to eigenvalue change.

Description

Milling cutter state on_line monitoring and method for early warning during complex surface machining

Technical field

During Impeller Machining of the present invention, the method for milling cutter state on_line monitoring belongs to field of diagnosis about equipment fault, Relate to selection three aspects of calculating of the selection monitoring signal kinds and process, the extraction of characteristic information and characteristic quantity.

Background technology

Along with the large-scale precision machinery such as aero-engine, compression is day by day to high load capacity, high efficiency and high reliability side To development, the suface processing quality of the critical component-impeller in them is required more and more higher.Owing to spoon of blade structure is multiple Miscellaneous, the reason such as the narrow space between blade, in the course of processing, knife bar needs to use bigger Mold processing, and this easily causes cutter Milling cutter generation tipping fixing bottom bar, abrasion, disrepair phenomenon.

The suface processing quality of blade and milling cutter state have substantial connection, tipping, abrasion, damaged milling cutter and blade surface Between frictional force can be increased dramatically, cutting temperature can steeply rise, the crudy of blade surface can be remarkably decreased, therefore blade In the course of processing, the on-line monitoring of milling cutter state has very important reality meaning to the raising crudy of blade, working (machining) efficiency Justice.

Good fortune occasion in Japanese scholars village detect based on cutting component than reflection cutter mill at document cutting component ratio injury Damage situation, but force transducer exists, and installation is inconvenient, loaded area is little, high in cost of production unfavorable factor, it is impossible to it is applied to actual prison Survey.Document dynamic milling based on vibration cutting displacement force measuring method points out that milling cutter can be by interrupted cut in the course of processing The excitation of power, driving frequency be cutter tooth by frequency, when the speed of mainshaft is relatively low, driving frequency be far below tooling system first Rank natural frequency, it is believed that the stiffness characteristics of cutter is essentially identical with time static, i.e. milling cutter are by after Milling Force effect being elasticity Deformation, the vibration displacement of the dynamic milling force on milling cutter and surveyed handle of a knife meets linear relationship, therefore can by measurement physical quantity from The measurement of Milling Force is converted to the measurement of handle of a knife vibration displacement, and the present invention is demonstrating Milling Force and handle of a knife swing current vortex displacement On the basis of there is linear corresponding relation between signal, it is considered to utilize current vortex displacement signal to replace force signal as monitoring letter Number, but because current vortex sensor there is also, inconvenient problem is installed, then propose to be added by collection lathe spindle nose X, Y-direction Rate signal, then to acceleration signal secondary frequency volume integration obtain displacement signal, extract integral displacement signal fundamental frequency and Harmonic signal, as monitoring signal, on the one hand solves sensor and installs the problem of inconvenience, effectively eliminate interference signal simultaneously Impact.

Summary of the invention

The present invention is demonstrating on the basis of Milling Force and handle of a knife swing and there is linear relationship between current vortex displacement signal, The machine tool chief axis end X that records with three-dimensional acceleration transducer, the acceleration signal of Y-direction, to its secondary frequency volume integration conduct Handle of a knife displacement signal, overcomes all drawbacks directly being carried out displacement detecting in Machining of Curved Surface by eddy current displacement sensor, On the basis of the handle of a knife vibration displacement signal using Synchronous time average the Cumulant Method Using to go out integration goes interference to process, carry Go out using the fundamental frequency in integral displacement signal and harmonic signal thereof as monitoring signal, solved milling during complex surface machining Cutter state on_line monitoring and an early warning difficult problem.

Technical scheme:

Milling cutter state on_line monitoring and method for early warning during a kind of complex surface machining, step is as follows:

Step A. obtains the acceleration signal stage online: three-dimensional acceleration transducer is fixed on machine tool chief axis end, mark Determine to measure direction to three-dimensional acceleration transducer to be completely superposed with milling handle vibration X-direction and Y-direction, three-dimensional acceleration sensing Device is for measuring milling handle vibration X-direction and the acceleration signal of Y-direction；

Step B. acceleration signal is converted to monitor signal phase: the milling handle that step A obtains vibrates X-direction and Y side To acceleration signal, monitoring signal be milling handle vibration displacement signal, to step A obtain milling handle vibration X side Carry out quadratic integral to the acceleration signal with Y-direction and be converted into displacement signal, represent milling handle vibration displacement；Owing to accelerating Degree signal band wide ranges, the signal component comprised is complex, causes integral displacement signal to be mixed with more interference signal, in order to Remove the impact of interference signal, after utilizing Fourier transform (FFT, Fast Fourier Transformation) to integration Displacement signal carries out spectrum analysis, filters other frequency signals and only retains fundamental frequency and harmonic components thereof, and the inverse Fourier of recycling becomes Change (IFFT, Inverse Fast Fourier Transformation) and become filtered signal again time-domain signal as prison Survey signal, in order to verify the effectiveness of monitoring signal measurement vibration cutting displacement, to current vortex displacement signal and monitoring signal pair The cumulant answered carries out cross-correlation analysis, using the fundamental frequency of integral displacement signal and harmonic signal thereof as monitoring signal, on the one hand Solve sensor and the problem of inconvenience is installed, can effectively remove the impact of interference signal simultaneously.

The step C. monitoring signal online treatment stage: monitoring milling handle vibrated by Synchronous time average the Cumulant Method Using Signal is further processed, and strengthens the reliability and stability of monitoring signal.

Synchronous time average the Cumulant Method Using principle is as follows: the periodic waveform of vibration cutting displacement signal is close with cutting tool state Relevant, during Impeller Machining, working conditions change causes monitoring signal waveform amplitude and changes greatly in different time sections, time domain Synchronized averaging can be with the effective attenuation frequency signal unrelated with gyration period, therefore, monitoring signal is carried out Synchronous time average Can strengthen the stability of monitoring signal period waveform, this is extremely important to relying on signal waveform to carry out feature extraction；Time domain is same Step average accumulated amount is the overall summary to signal, in order to describe the integrality of signal and extract signal letter from different perspectives Breath, uses Synchronous time average cumulant to be described monitoring signal, and compared with primary signal, the periodic waveform of cumulant is more Add stable, thus monitoring result is relatively reliable.

It is as follows that Synchronous time average the Cumulant Method Using is treated journey to milling handle vibration monitoring signal: sets rotating machinery Vibration signal be x_t, speed is f₀, the sampling interval is Δ, and the discrete signal that vibration signal is corresponding is x_t=x (n Δ), will Vibration signal is according to speed f₀Extract corresponding vibration signal, if by x_nIntercept as P section, then the cycle of each period isAssume that the sampling number of every section is equal and value is N, synchronized averagingJust can represent with following formula:

${\stackrel{\‾}{x}}_n=\frac{1}{P}\underset{P=0}{\overset{P-1}{\Σ}}{x}_{n+PN}---\left(1\right)$

Algorithm principle based on Synchronous time average, in conjunction with the computing formula of statistics cumulant, derives Domain Synchronous The computing formula of average accumulated amount, the single order derived, second order Synchronous time average cumulant formula are as follows:

Single order cumulant:

Second-order cumulant:

Step D. cutting tool state judges the stage online: on the basis of step C, is tired out Synchronous time average by characteristic quantity Accumulated amount waveform is quantitatively described, it is judged that cutting tool state.

Before and after milling cutter breakage, there is the biggest difference in the signal waveform of the Synchronous time average cumulant of handle of a knife vibration displacement, Utilizing time domain index quantitatively to portray cumulant waveform, the different corresponding time domain index value of waveform certainly exists difference, by time domain index As characteristic quantity, realize tool condition monitoring according to the difference of desired value before and after tool failure, use variance, degree of skewness, kurtosis Value, absolute mean and virtual value portray signal waveform as characteristic quantity, and wherein, variance is for describing the discrete journey of signal waveform Degree, degree of skewness is for describing skew direction and the degree of data, and kurtosis value is especially sensitive to impact signal, and absolute mean embodies whole Body amplitude size, feature extraction thinking is as shown in flow chart 1, and each time domain index correspondence formula is as follows:

Variance:

Degree of skewness:

Kurtosis value:

Absolute value average:

Virtual value:

Wherein y_iRepresent single order, second order sync average accumulated amount,For single order, second order sync average accumulated amount average.

Beneficial effects of the present invention: the present invention utilizes acceleration transducer to gather signal, and acceleration transducer has installation Freely measure the feature of simplicity, can be used for actual monitoring；Synchronized averaging cumulant is utilized to extract the characteristic information of milling cutter state, then In conjunction with characteristic quantity the feature extracting method that synchronized averaging cumulant is quantitatively described had monitoring result is good, programming letter Feature single, that the speed of service is fast, meets the requirement of on-line monitoring real-time.

Accompanying drawing explanation

Fig. 1 is milling cutter status monitoring flow chart.

Fig. 2 a is X-direction force signal and current vortex displacement linearly relation displacement diagram.

Fig. 2 b is Y-direction force signal and current vortex displacement linearly relation displacement diagram.

Fig. 3 is point of a knife to frequency response function figure at displacement measurement.

Fig. 4 a is the 14th group of data current vortex displacement signal single order accumulation spirogram.

Fig. 4 b is the 14th group of data current vortex displacement signal second-order cumulant figure.

Fig. 4 c is the 14th group of data integral displacement signal single order accumulation spirogram.

Fig. 4 d is the 14th group of data integral displacement signal second-order cumulant figure.

Fig. 5 a is that the 14th group of data current vortex displacement signal returns scatterplot with integral displacement signal single order cumulant linearly Figure.

Fig. 5 b is that the 14th group of data current vortex displacement signal returns scatterplot with integral displacement signal second-order cumulant linearly Figure.

Fig. 6 a is the 64th group of data current vortex displacement signal single order accumulation spirogram.

Fig. 6 b is the 64th group of data current vortex displacement signal second-order cumulant figure.

Fig. 6 c is the 64th group of data integral displacement signal single order accumulation spirogram.

Fig. 6 d is the 64th group of data integral displacement signal second-order cumulant figure.

Fig. 7 a is that the 64th group of data current vortex displacement signal returns scatterplot with integral displacement signal single order cumulant linearly Figure.

Fig. 7 b is that the 64th group of data current vortex displacement signal returns scatterplot with integral displacement signal second-order cumulant linearly Figure.

Fig. 8 a is 1-80 group data single order cumulant absolute value average figure.

Fig. 8 b is 1-80 group data single order cumulant variance yields figure.

Fig. 8 c is 1-80 group data single order cumulant virtual value figure.

Fig. 8 d is 1-80 group data single order cumulant kurtosis value figure.

Fig. 8 e is 1-80 group data single order cumulant deflection angle value figure.

Fig. 9 a is 1-80 group data second-order cumulant absolute value average figure.

Fig. 9 b is 1-80 group data second-order cumulant variance yields figure.

Fig. 9 c is 1-80 group data second-order cumulant virtual value figure.

Fig. 9 d is 1-80 group data second-order cumulant kurtosis value figure.

Fig. 9 e is 1-80 group data second-order cumulant deflection angle value figure.

Detailed description of the invention

Below in conjunction with technical scheme and accompanying drawing, further illustrate the detailed description of the invention of the present invention.

Experimental data totally 80 groups, is collected in certain enterprise's compressor impeller Milling Process production scene domestic, in gatherer process, Shutting down at interval of a period of time and observe milling cutter state, find the 61st group of milling cutter tipping, experiment parameter is as shown in table 1.

Table 1 experiment parameter

14th group, the 64th group of data be respectively the data before and after milling cutter tipping, the synchronized averaging cycle is 10 speed cycles, Fig. 4 a, Fig. 4 b the most corresponding 14th group of data current vortex displacement signal single order, the result of calculation of second-order cumulant, Fig. 4 c, Fig. 4 d The most corresponding 14th group of integral displacement signal single order, the result of calculation of second-order cumulant；The most corresponding 64th group of number of Fig. 6 a, Fig. 6 b According to current vortex displacement signal single order, the result of calculation of second-order cumulant；The most corresponding 64th group of integral displacement signal of Fig. 6 c, Fig. 6 d Single order, the result of calculation of second-order cumulant.

In order to describe the wave-form similarity between current vortex displacement signal and integral displacement signal, calculate the 14th group and the 64th Correlation coefficient between group data current vortex displacement signal cumulant corresponding with integral displacement signal, and draw scatterplot, Fig. 5 a, figure The most corresponding regression result between the 14th group of data two class displacement signal single order, second-order cumulant of 5b, Fig. 7 a, Fig. 7 b correspondence respectively Regression result between the 64th group of data two class displacement signal single order, second-order cumulant.

Found out that correlation coefficient all reaches more than 0.9 by Fig. 5 a, Fig. 5 b and Fig. 7 a, Fig. 7 b, show current vortex displacement signal With integral displacement signal linearly relation highly significant, both have similar monitoring effect, it is possible to use integral displacement signal Substitute current vortex displacement signal and carry out tool condition monitoring；Corresponding comparison diagram 4a, Fig. 4 b, Fig. 4 c, Fig. 4 d Yu Fig. 6 a, Fig. 6 b, figure 6c, Fig. 6 d finds out, after tool failure, the single order of two class displacement signals, second-order cumulant amplitude all increased, the most all with two The amplitude of rank cumulant increases notable；Further, it is also possible to find out that the second-order cumulant amplitude of integral displacement signal increases than electric whirlpool Stream displacement signal becomes apparent from, and this shows the second-order cumulant second-order cumulant than current vortex displacement signal of integral displacement signal Sensitiveer to cutting tool state reacting condition.

Fig. 8 a, Fig. 8 b, Fig. 8 c, Fig. 8 d, Fig. 8 e and Fig. 9 a, Fig. 9 b, Fig. 9 c, Fig. 9 d, Fig. 9 e represent that integral displacement is believed respectively Number single order, the feature extraction result of second-order cumulant, found out integral displacement signal by Fig. 8 a, Fig. 8 b, Fig. 8 c, Fig. 8 d, Fig. 8 e The each characterizing magnitudes of single order cumulant entirety after the 55th group increases, and Fig. 9 a, Fig. 9 b, Fig. 9 c, Fig. 9 d, Fig. 9 e will become apparent from The each characterizing magnitudes of single order cumulant of integral displacement signal increases notable at the 55th group, and eigenvalue is basicly stable afterwards, Liang Zhejie Conjunction can be determined that tool failure occurs at the 55th group, and experimental record tool failure occurs at the 61st group, and this monitoring result is in the Before 61 groups, thus monitoring result meets the requirements.

Claims (1)

1. milling cutter state on_line monitoring and method for early warning during a complex surface machining, it is characterised in that step is as follows:

Step A. obtains the acceleration signal stage online: three-dimensional acceleration transducer is fixed on machine tool chief axis end, demarcates extremely Three-dimensional acceleration transducer is measured direction and is completely superposed with milling handle vibration X-direction and Y-direction, and three-dimensional acceleration transducer is used In measuring milling handle vibration X-direction and the acceleration signal of Y-direction；

Step B. acceleration signal is converted to monitor signal phase: milling handle vibration X-direction that step A obtains and Y-direction Acceleration signal, monitoring signal is the displacement signal of milling handle vibration, the milling handle vibration X-direction obtaining step A and Y The acceleration signal in direction carries out quadratic integral and is converted into displacement signal, represents milling handle vibration displacement；Fourier is utilized to become Change FFT and displacement signal is carried out spectrum analysis, filter other frequency signals and only retain fundamental frequency and harmonic components thereof, inverse Fu of recycling Filtered signal is become again time-domain signal as monitoring signal by vertical leaf transformation IFFT；

The step C. monitoring signal online treatment stage: monitoring signal milling handle vibrated by Synchronous time average the Cumulant Method Using It is further processed, strengthens the reliability and stability of monitoring signal；

It is as follows that Synchronous time average the Cumulant Method Using is treated journey to milling handle vibration monitoring signal: sets rotating machinery and shakes Dynamic signal is x_t, speed is f₀, the sampling interval is Δ, and the discrete signal that vibration signal is corresponding is x_t=x (n Δ), will vibration Signal is according to speed f₀Extract corresponding vibration signal, if by x_nIntercept as P section, then the cycle of each period is Assume that the sampling number of every section is equal and value is N, synchronized averagingRepresent with following formula:

${\stackrel{\‾}{x}}_n=\frac{1}{P}\underset{P=0}{\overset{P-1}{\Σ}}{x}_{n+PN}---\left(1\right)$

Algorithm principle based on Synchronous time average, in conjunction with the computing formula of statistics cumulant, derives Synchronous time average The computing formula of cumulant, the single order derived, second order Synchronous time average cumulant formula are as follows:

Single order cumulant:

Second-order cumulant:

Step D. cutting tool state judges the stage online: before and after milling cutter breakage, the Synchronous time average cumulant of handle of a knife vibration displacement Signal waveform there is the biggest difference, utilize time domain index quantitatively to portray cumulant waveform, the different corresponding time domain index value of waveform There are differences, using time domain index as characteristic quantity, realize tool condition monitoring according to the difference of desired value before and after tool failure, adopt Portraying signal waveform by variance, degree of skewness, kurtosis value, absolute mean and virtual value as characteristic quantity, wherein, variance is used for describing The dispersion degree of signal waveform, degree of skewness is for describing skew direction and the degree of data, and kurtosis value is the quickest to impact signal Sense, absolute mean embodies overall amplitude size, and each time domain index correspondence formula is as follows:

Variance:

Degree of skewness:

Kurtosis value:

Absolute value average:

Virtual value:

Wherein y_iRepresent single order, second order sync average accumulated amount,For single order, second order sync average accumulated amount average.