CN108247163B - A method of establishing wire cutting power curve mutation waveform characterization libraries - Google Patents
A method of establishing wire cutting power curve mutation waveform characterization libraries Download PDFInfo
- Publication number
- CN108247163B CN108247163B CN201611232737.1A CN201611232737A CN108247163B CN 108247163 B CN108247163 B CN 108247163B CN 201611232737 A CN201611232737 A CN 201611232737A CN 108247163 B CN108247163 B CN 108247163B
- Authority
- CN
- China
- Prior art keywords
- mutation
- wave band
- waveform
- feature vector
- amplitude
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23H—WORKING OF METAL BY THE ACTION OF A HIGH CONCENTRATION OF ELECTRIC CURRENT ON A WORKPIECE USING AN ELECTRODE WHICH TAKES THE PLACE OF A TOOL; SUCH WORKING COMBINED WITH OTHER FORMS OF WORKING OF METAL
- B23H11/00—Auxiliary apparatus or details, not otherwise provided for
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23H—WORKING OF METAL BY THE ACTION OF A HIGH CONCENTRATION OF ELECTRIC CURRENT ON A WORKPIECE USING AN ELECTRODE WHICH TAKES THE PLACE OF A TOOL; SUCH WORKING COMBINED WITH OTHER FORMS OF WORKING OF METAL
- B23H2600/00—Machining conditions
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Electrical Discharge Machining, Electrochemical Machining, And Combined Machining (AREA)
Abstract
The present invention provides a kind of methods for establishing wire cutting power curve key point wave band identification database, include the following steps: 1. opening wire cutting machine tool control cabinet host computers, open bath;2. keeping washing by water and opening wire transporting;3. keeping bath and wire transporting;4. opening the cutting of lathe air transportion silk;5. limit block gap is turned down, dry running cutting is done;6. cutting 1mm stainless steel plate;7. cutting workpiece, limit block gap is adjusted to minimum;8. cutting vertical duction after workpiece;9. straight cut 1mm stainless steel plate backarc wire cutting;10. carrying out θ angle after incision turns to cutting;11. changing cutting workpiece is n millimeters;12. changing workpiece material is other materials;13. vector, which is arranged, is superimposed coefficient;14. self study differentiates typing new feature vector;15. correcting edge data;16. establishing database, data are imported.As described above, the present invention can establish line cutting processing mutation power curve feature database by the method, to provide convenience for the machine power based on this feature database, energy consumption correlative study.
Description
[technical field]
The present invention relates to a kind of methods for establishing wire cutting power curve mutation waveform characterization libraries, more particularly to a kind of line
The self-learning method of cutting power curve mutation waveform characterization libraries.
[background technique]
Wire cutting is discharged using high-frequency impulse, has working solution filling between gap, loads cathode, workpiece in wire electrode at this time
Load anode;Cathode runaway electron, electronics faces south under electric field acceleration hits the neutral particle of working solution during Ghandler motion moves, in
Property particle generate electronics and cation under shock, the ion and electronics of generation are added again in knockout process, and avalanche of ionization formula is sent out
Exhibition.It is ionized breakdown in working solution, electronics faces south, and Ghandler motion is dynamic, and ion forms discharge channel to movable cathode, generates spark and puts
Electricity, can form 10000 degrees Celsius or more of high temperature, and instant melting even gasifies workpiece material, simultaneously as this process pole
Short, gasification has explosive effect, and the power of generation is dished out the workpiece material dissolved, and the material being then melted is in electrolyte
Middle regelation is drained at small sphere, and by electrolyte.
In order to understand the changed power of each different phase of wire cutting when processing the workpiece in depth, this just needs to cut on line
Each stage boundary when cutting processing is found out, in order to accurately correspond to the power curve in each stage, to specify
Solve the power situation in each process segment.But manually the methods of judgement carrys out the boundary of each stage of identification line cutting power curve
Place, with this different phase for distinguishing whole process be extremely inaccurate and workload is huge.Additionally by lathe
The NC code of control system can also accurately find out the boundary between each process segment, but the control program of lathe band is with guarantor
Protection mechanism, NC code are not easy to obtain and carry out data preservation.Therefore, in order to distinguish the power curve of entire line cutting processing
Boundary needs to establish waveform characterization libraries possessed by the power curve of a linear cutter stage boundary.By wire cutting
Each process segment boundary corresponds to the upper mutation waveform of power curve, and the Primary mutations wave band of wire cutting power curve has
Mutation waveform is opened in bath, wire transporting opens mutation waveform, silk cylinder turns to mutation waveform, opens cutting mutation waveform, incision mutation
Waveform, cutting turn to mutation waveform.To the mutation waveform extracting mean value p of acquisitionj, amplitude pm, pole line slope k, for mutation
Waveform is overlapped the addition of waveforms generated, needs to be arranged superposition coefficient, for calculating, correcting the feature vector of superposition.It is adopted with experiment
The characteristic vector data of collection makees sample, and data precision is improved by way of machine learning, increases data volume, establishes huge base
Plinth database.Power curve of the existing a variety of data acquisition devices to obtain lathe, it is bent using the power of data acquisition device
Line obtains each basic data with experimental method.
[summary of the invention]
The object of the invention is to solve the problems of the prior art, propose that a kind of wire cutting power curve of establishing is mutated wave
The method of shape feature database can be mutated waveform characterization libraries by establishing this wire cutting power curve, using characteristic wave bands sample, more
The chugging situation of each processing node in good understanding process, thus crucial for the power curve based on this database
Point recognition methods provides support, also provides basis for follow-up study.
To achieve the above object, the present invention proposes a kind of method for establishing wire cutting power curve mutation waveform characterization libraries,
Method includes the following steps:
Step 1: opening wire cutting machine tool control cabinet host computer, power curve is obtained by power collecting device, to function
After rate is steady, bath is opened, extracts the feature vector of power curve mutation waveform.
Step 2: keeping bath, wire transporting is steadily opened afterwards in power curve, extracts caused mutation waveform when opening wire transporting
Feature vector.
Step 3: adjusting lathe two limits block gap to the minimum range allowed, bath is opened, wire transporting is carried out, to power
Curve occurs stable continuous cycles mutation with not stalling to for silk cylinder, analyzes the waveform in each period, extracts this
The feature vector of mutation waveform in power waveform.
Step 4: adjusting limited block distance to normal, bath, wire transporting are all closed, and when main engine power is steady, are separately operable
Lathe does the cutting of air transportion silk along x, y-axis direction, extracts mutation wave band, calculates the feature vector of this mutation waveform.
Step 5: starting lathe does straight line dry running along x, y-axis respectively and cuts after limited block distance is adjusted to allow minimum value
It cuts, according to periodic breaks power curve at this time, intercepts the mutation wave band in complete cycle, calculate the feature of this mutation waveform
Vector.
Step 6: limited block distance is adjusted to the degree that wire electrode can be made to cover as far as possible, lathe is separately operable along x, y-axis
1mm meters of stainless steel plates of straight cut extract the feature vector of mutation waveform when incision.
Step 7: limited block distance is adjusted to minimum, function is obtained after wire electrode straight line cuts 1mm stainless steel work-piece along x, y
Periodic breaks section in rate curve, the mutation waveshape feature vector of interception standard the most.
Step 8: limited block distance is adjusted to the size degree that wire electrode can be made to cover as far as possible, respectively along x, y straight line
After cutting 1mm stainless steel work-piece, it is carried out continuously vertical duction cutting, this is extracted and turns to corresponding sudden change sample wave band, calculate
It is mutated the feature vector of waveform.
Step 9: carrying out curvature is respectively after x, y straight cut 1mm stainless steel work-pieceArc cutting, extraction cuts
Secant routing straight line is changed into corresponding wave band when curve, calculates the feature vector of this mutation waveform.
Step 10: turning to respectively after x, y straight cut 1mm stainless steel work-piece and making to cut with the oblique line of this straight line degree into θ angle
It cuts, corresponding power curve is mutated wave band when extracting cutting route turnover, calculates the feature vector of mutation waveform.
Step 11: changing 1 millimeter of stainless steel work-piece into n millimeters of stainless steel work-pieces, the six, the seven, eight, nine, ten steps are repeated,
It can extract each strategic point mutations wave band under different-thickness, calculate the feature vector of mutation waveform.
Step 12: stainless steel work-piece is changed into other materials workpiece, repeat the six, the seven, eight, nine, ten, 11
Step, can extract each key point wave band under the corresponding different-thickness of different materials workpiece, calculate mutation waveform feature to
Amount.
Step 13: setting superposition coefficient η, then obtain different wave mutation addition of waveforms caused by synchronization and generate
Wave band feature vector.
Step 14: self study typing new feature vector, can with support vector machines (Support Vector Machine,
Abbreviation SVM) setting self study mode.
Step 15: amendment edge data.
Step 16: establishing database, all of above sample vector is imported into database according to key point category classification.
Preferably, the bath is mutated wave band mean value pjsWith amplitude pms, wire transporting be mutated wave band mean value pjsyWith amplitude
pmsy, bath wire transporting open when silk cylinder turn to mutation wave band mean value pjsyzWith amplitude pmsyz, open air transportion silk cutting mutation wave band it is equal
Value pjkWith amplitude pmk, air transportion silk cutting wire cylinder turn to mutation wave band mean value pjkzWith amplitude pmkz, incision steel plate be mutated wave band mean value
pjrWith amplitude pmr, silk cylinder turns to mutation wave band mean value p after incisionjrzWith amplitude pmrz, vertical duction be mutated wave band mean value pjrcWith
Amplitude pmrc, arc cutting mutation wave band mean value pjrqWith amplitude pjrq, angle cutting mutation wave band mean value pjrzWith amplitude pmrz, change
Wave band mean value p is mutated for n millimeters of work piece cutsjhWith amplitude pmh, change other materials mutation wave band mean value pjcWith amplitude pmc, superposition
Wave band mean value is η1(pj1+pj2) with amplitude be η2(pm1+pm2), new feature vector mean value pjxWith amplitude pmx, threshold epsilon, ε1、ε2、ε3
Unit be w;Bath mutation wave band pole line slope ks, wire transporting be mutated wave band pole line slope ksy, bath wire transporting open
Silk cylinder turns to mutation wave band pole line slope k when openingsyz, open air transportion silk cutting mutation wave band pole line slope kk, air transportion
Silk cutting wire cylinder turns to mutation two pole line slope k of wave bandkz, incision steel plate be mutated two pole line k of wave bandr, silk after incision
Cylinder turns to mutation wave band pole line distance krz, vertical duction be mutated wave band pole line slope krc, arc cutting mutation wave band
Two pole line slope ksrq, angle cutting mutation wave band pole line slope krz, be changed to n millimeters of work piece cuts mutation wave band poles
Value point line slope kh, other materials mutation wave band extreme point line slope be η3(k1+k2), superposition coefficient η1、η2、η3It is nothing
Dimensions number.
Beneficial effects of the present invention: the present invention can establish line cutting processing each process segment by the method
The power curve of interface point is mutated waveform characterization libraries, to provide for the machine power based on this feature database, energy consumption correlative study
It is convenient.
[Detailed description of the invention]
Fig. 1 is the method and step figure that the present invention establishes wire cutting power curve mutation waveform characterization libraries.
Power curve mutation waveform diagram when Fig. 2 is wire cutting machine tool incision workpiece.
[specific embodiment]
The present invention is further detailed with reference to the accompanying drawings and detailed description.
As shown in Figure 1, the present invention provides a kind of methods for establishing wire cutting power curve mutation waveform characterization libraries.
This feature library method for building up includes the following steps:
Step 1: opening wire cutting machine tool control cabinet host computer, power curve is obtained by power collecting device, to function
After rate is steady, bath is opened, extracts the feature vector of power curve mutation waveform.
In the step, the vector of extraction are as follows: mean valueWherein t1And t2Distinguish wave band initial time, end thus
Only time, p indicate wire cutting power curve function;AmplitudeWherein pmaxWith pminWave band thus respectively
Greatly, minimum;Two pole line slopes:Wherein tmaxAnd tminRespectively wire cutting power curve exists
Maximum p in this wave bandmaxWith minimum pminCorresponding time point.
Step 2: keeping bath, wire transporting is steadily opened afterwards in power curve, extracts caused mutation waveform when opening wire transporting
Feature vector.
In the step, obtains wave band and calculated three-dimensional samples vector is respectively as follows: mean value pjsyWith amplitude pmsyAnd two
A pole line slope ksy。
Step 3: adjusting lathe two limits block gap to the minimum range allowed, bath is opened, wire transporting is carried out, to power
Curve occurs stable continuous cycles mutation with not stalling to for silk cylinder, analyzes the waveform in each period, extracts this
The feature vector of mutation waveform in power waveform.
In the step, extracts the feature vector in this power waveform and be respectively as follows: mean value pjsyz, amplitude pmsyzAnd two poles
The slope k of point linesyz, as only open under bath and wire transporting state, it is corresponding that wire transporting cylinder turns to the key point wave band generated
Three-dimensional samples vector.
Step 4: adjusting limited block distance to normal, bath, wire transporting are all closed, and when main engine power is steady, are separately operable
Lathe does the cutting of air transportion silk along x, y-axis direction, extracts mutation waveform, calculates the feature vector of this mutation waveform.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjk, amplitude pmkAnd two pole line slope
kk, corresponding mutation wave band sample vector when obtaining lathe bath, wire transporting, spindle motor while opening.
Step 5: starting lathe does straight line dry running along x, y-axis respectively and cuts after limited block distance is adjusted to allow minimum value
It cuts, according to periodic breaks power curve at this time, intercepts the mutation wave band in complete cycle, calculate the feature of this mutation waveform
Vector.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjkz;Amplitude pmkz;Two pole line slope kskz,
As corresponding silk cylinder steering causes the corresponding sample vector of chugging before lathe incision.
Step 6: limited block distance is adjusted to the degree that wire electrode can be made to cover as far as possible, lathe is separately operable along x, y-axis
1mm meters of stainless steel plates of straight cut extract the feature vector of mutation waveform when incision.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjrWith amplitude pmrWith two pole line krTo get
To the sample vector of mutation wave band.
Step 7: limited block distance is adjusted to minimum, function is obtained after wire electrode straight line cuts 1mm stainless steel work-piece along x, y
Periodic breaks section in rate curve, the mutation waveshape feature vector of interception standard the most.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjrh, amplitude pmrh, pole line distance krh, i.e.,
For after this workpiece of wire electrode straight cut, silk cylinder turns to the three-dimensional feature vector of corresponding mutation wave band.
Step 8: limited block distance is adjusted to the size degree that wire electrode can be made to cover as far as possible, respectively along x, y straight line
After cutting 1mm stainless steel work-piece, it is carried out continuously vertical duction cutting, this is extracted and turns to corresponding sudden change sample wave band, calculate
It is mutated the feature vector of waveform.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjrc, amplitude pmrc, and two pole lines are oblique
Rate krc, it is corresponding prominent by running to this process of another electric motor starting simultaneously that stops to obtain x, a motor in y-axis motor
Become the three-dimensional feature vector of wave band.
Step 9: carrying out curvature is respectively after x, y straight cut 1mm stainless steel work-pieceArc cutting, extraction cuts
Secant routing straight line is changed into corresponding wave band when curve, calculates the feature vector of this mutation waveform.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjrq, amplitude pmrqAnd two pole line it is oblique
Rate krq, obtain x, y motor and meanwhile starting but with the mutation band three-dimensional feature vector of friction speed synthetic operation.
Step 10: turning to respectively after x, y straight cut 1mm stainless steel work-piece and making to cut with the oblique line of this straight line degree into θ angle
It cuts, corresponding power curve is mutated wave band when extracting cutting route turnover, calculates the feature vector of mutation waveform.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjrz, amplitude pmrz, and two pole lines are oblique
Rate krz, up to the corresponding three-dimensional feature vector of the angled turnover of wire electrode cutting path.
Step 11: changing 1 millimeter of stainless steel work-piece into n millimeters of stainless steel work-pieces, the six, the seven, eight, nine, ten steps are repeated,
It can extract each strategic point mutations wave band under different-thickness, calculate the feature vector of mutation waveform.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjh, amplitude pmhAnd extreme point line slope
kh, form corresponding three-dimensional feature vector.
Step 12: stainless steel work-piece is changed into other materials workpiece, repeat the six, the seven, eight, nine, ten, 11
Step, can extract each key point wave band under the corresponding different-thickness of different materials workpiece, calculate mutation waveform feature to
Amount.
In the step, the feature vector for calculating this wave band is respectively as follows: mean value pjc, amplitude pmc, extreme point line slope kc, group
At corresponding three-dimensional feature vector.
Step 13: setting superposition coefficient η, then obtain different wave mutation addition of waveforms caused by synchronization and generate
Wave band feature vector.
In the step, the feature vector for the wave band that addition of waveforms caused by waveform difference critical event synchronization generates
Mean value is η1(pj1+pj2);Amplitude is η2(pm1+pm2);Two extreme point line slopes are η3(k1+k2), wherein η1、η2、η3Respectively
Two different key points are overlapped hourly values, amplitude, superposition coefficient corresponding to extreme point line slope, pj1With pj2、pm1With pm2、
k1With k2Respectively two different key points correspond to the mean value, amplitude and extreme point line slope of wave band.
Step 14: self study typing new feature vector, can use support vector machines (Support Vector
Machine, abbreviation SVM are a kind of mode identification methods based on Statistical Learning Theory) setting self study mode.
In the step, the method for setting support vector machines self study mode is: the feature vector obtained with front, as ginseng
This, after the new power curve waveform of typing, calculates the feature vector mean value p of new waveform in the same old wayjx, amplitude pmx, extreme point line
Slope kx, feature vector and sampling feature vectors are subjected to preliminary comparison according to formula (1), if when formula (1) is set up, then it is assumed that new
Wave band feature vector is classified as the corresponding wave band of this kind of key points, then is with formula (2) with new feature vector by sampling feature vectors
According to fine correlation is made, if formula (2) is all set up, i.e., this wave band is determined as such feature vector,
In formula (1), ε is similarity given threshold;In formula (2), ε1、ε2、ε3Respectively amplitude pj, mean value pm, extreme point connect
The threshold value of line slope K.
Step 15: amendment edge data.
The step is to correct the data recording error that may be present based on preceding step, and support vector machines is extracted prominent
When becoming wave band, judge that the identification process of classification has deviation, in order to guarantee the accuracy of new logging data, first not at once by new feature
Vector is entered into database, but new feature vector is set as sample vector to be recorded.Artificial correction sample vector corresponding sides to be recorded
Edge data will differentiate ambiguous data, for instance it can be possible that the wave band that silk cylinder turns to may be wire electrode incision workpiece again
Wave band, progress are manually modified classification by actual conditions.
Step 16: establishing database, all of above feature vector is imported into data according to the classification of type of mutation waveform
Library.
In the step, the method for establishing database is: firstly, establishing workpiece material thickness table, which includes " workpiece material
Material ", " thickness of workpiece ", " material _ thickness _ ID " three parameters, wherein " material _ thickness _ ID " is by " workpiece material " and " workpiece is thick
Degree " determines;Then, mutation type of waveform table is established, which includes " mutation type of waveform _ ID " and " material _ thickness _ ID " two
Parameter, " material _ thickness _ ID " in the mutation type of waveform table in " material _ thickness _ ID " and workpiece material thickness table keep one
It causes;Finally, establish mutation wave character vector storage table, the table include " mutation Band Class _ ID ", " mutation wave band feature to
Amount ", " acquisition time " three parameters, the 12nd step treated mutation Waveform storage is in mutation wave character vector storage table
In " mutation Band Class _ ID ", the feature vector for the mutation waveform that the 13rd step is extracted is stored in mutation wave character vector and deposits
It stores up in " the mutation wave band feature vector " of table, " mutation Band Class _ ID " and the mutation wave of mutation wave character vector storage table
" the mutation Band Class _ ID " of shape type list is consistent.
The step of being mutated waveform using the data base querying are as follows: " workpiece material " and " workpiece thickness in workpiece material thickness table
Degree " determines unique " material _ thickness _ ID ";Utilize " material _ thickness _ ID " material can be found in mutation type of waveform table
The mutation waveform position stored under thickness, i.e. " mutation type of waveform _ ID ";Recycle " mutation type of waveform _ ID " in mutation wave
The eigenvector information met the requirements is found in shape feature vector storage table.This database stores workpiece information (material, thickness simultaneously
Degree) and mutation shape information (type of waveform, feature vector, acquisition time).
SQL (full name in English: Structured Query Language) data can be used in database in 14th step
DB2 also can be used in library (DB2 database is a kind of distributed data base solution DATABASE 2 that IBM Corporation announces)
Database.
Specifically, the bath is mutated wave band mean value pjsWith amplitude pms, wire transporting be mutated wave band mean value pjsyWith amplitude pmsy、
Silk cylinder turns to mutation wave band mean value p when wire transporting of washing by water is openedjsyzWith amplitude pmsyz, open air transportion silk cutting mutation wave band mean value pjk
With amplitude pmk, air transportion silk cutting wire cylinder turn to mutation wave band mean value pjkzWith amplitude pmkz, incision steel plate be mutated wave band mean value pjrWith
Amplitude pmr, silk cylinder turns to mutation wave band mean value p after incisionjrzWith amplitude pmrz, vertical duction be mutated wave band mean value pjrcWith amplitude
pmrc, arc cutting mutation wave band mean value pjrqWith amplitude pjrq, angle cutting mutation wave band mean value pjrzWith amplitude pmrz, be changed to n milli
Rice work piece cut is mutated wave band mean value pjhWith amplitude pmh, change other materials mutation wave band mean value pjcWith amplitude pmc, superposition wave band it is equal
Value is η1(pj1+pj2) with amplitude be η2(pm1+pm2), new feature vector mean value pjxWith amplitude pmx, threshold epsilon, ε1、ε2、ε3Unit
It is w;Bath mutation wave band pole line slope ks, wire transporting be mutated wave band pole line slope ksy, bath wire transporting open when silk
Cylinder turns to mutation wave band pole line slope ksyz, open air transportion silk cutting mutation wave band pole line slope kk, air transportion silk cutting
Silk cylinder turns to mutation two pole line slope k of wave bandkz, incision steel plate be mutated two pole line k of wave bandr, silk cylinder turns to after incision
It is mutated wave band pole line distance krz, vertical duction be mutated wave band pole line slope krc, arc cutting mutation two pole of wave band
Line slope krq, angle cutting mutation wave band pole line slope krz, be changed to n millimeters of work piece cuts mutation wave band extreme points and connect
Line slope kh, other materials mutation wave band extreme point line slope be η3(k1+k2), superposition coefficient η1、η2、η3It is dimensionless
Number.
The course of work of the present invention:
In i.e. a kind of course of work for the method for establishing wire cutting power curve mutation waveform characterization libraries of the present invention, foundation
Above-mentioned 16 obtain individual features vector suddenly step by step and summarize into basic database, wherein the mutation waveform obtained by experiment
Feature database is only applicable to same model lathe, and there are appropriate errors on the different lathes of same model, need to add correction factor
It is modified use.It is noted that adjusting limited block in third step, the 5th step, the 7th step apart from block to minimum value is allowed, it is
Mutation wave band is turned in order to obtain successional silk cylinder.And in other steps, the experimental period section measured must stablize to dash forward
After change, so limited block distance must be adjusted to permission maximum value, and also long wire electrode (such as: molybdenum filament) is loaded onto,
And when obtaining basic data, it is selected as the integral multiple that wire transporting cylinder turns to time half period, to reduce when obtaining data
It reduces silk cylinder and turns to mutation disturbance.
The present invention can establish the interface point power curve in each process segment in line cutting processing by the method
Waveform characterization libraries are mutated, to provide convenience for the machine power based on this feature database, energy consumption correlative study.
Above-described embodiment is the description of the invention, is not limitation of the invention, after any pair of simple transformation of the present invention
Scheme all belong to the scope of protection of the present invention.
Claims (6)
1. a kind of method for establishing wire cutting power curve mutation waveform characterization libraries, which is characterized in that this feature library method for building up
The following steps are included:
Step 1: opening wire cutting machine tool control cabinet host computer, power curve is obtained by power collecting device, it is flat to power
After steady, bath is opened, extracts the feature vector of power curve mutation waveform;
Step 2: keeping bath, wire transporting is steadily opened afterwards in power curve, extracts the spy of caused mutation waveform when opening wire transporting
Levy vector;
Step 3: adjusting lathe two limits block gap to the minimum range allowed, bath is opened, wire transporting is carried out, to power curve
Occur stable continuous cycles mutation with not stalling to for silk cylinder, analyzes the waveform in each period, extract this power
The feature vector of mutation waveform in waveform;
Step 4: adjusting limited block distance to normal, bath, wire transporting are all closed, and when main engine power is steady, are separately operable lathe
The cutting of air transportion silk is done along x, y-axis direction, extracts mutation wave band, calculates the feature vector of this mutation waveform;
Step 5: starting lathe does the cutting of straight line dry running along x, y-axis respectively after limited block distance is adjusted to allow minimum value,
According to periodic breaks power curve at this time, intercept the mutation wave band in complete cycle, calculate the feature of this mutation waveform to
Amount;Step 6: limited block distance is adjusted to the degree that wire electrode can be made to cover as far as possible, lathe is separately operable along x, y-axis straight line
1mm meters of stainless steel plates are cut, the feature vector of mutation waveform when incision is extracted;
Step 7: limited block distance is adjusted to minimum, it is bent to obtain power after wire electrode straight line cuts 1mm stainless steel work-piece along x, y
Periodic breaks section in line, the feature vector of the mutation waveform of interception standard the most;
Step 8: limited block distance is adjusted to the size degree that wire electrode can be made to cover as far as possible, respectively along x, y straight cut
After 1mm stainless steel work-piece, it is carried out continuously vertical duction cutting, this is extracted and turns to corresponding sudden change sample wave band, calculate mutation
The feature vector of waveform;
Step 9: carrying out curvature is respectively after x, y straight cut 1mm stainless steel work-pieceArc cutting, extract cutting line
Routing straight line is changed into corresponding wave band when curve, calculates the feature vector of this mutation waveform;
Step 10: turn to respectively after x, y straight cut 1mm stainless steel work-piece and make to cut with the oblique line of this straight line degree into θ angle,
Corresponding power curve is mutated wave band when extracting cutting route turnover, calculates the feature vector of mutation waveform;
Step 11: changing 1 millimeter of stainless steel work-piece into n millimeters of stainless steel work-pieces, the six, the seven, eight, nine, ten steps are repeated, can be mentioned
Each strategic point mutations wave band under different-thickness is got, the feature vector of mutation waveform is calculated;
Step 12: stainless steel work-piece is changed into other materials workpiece, repeat the six, the seven, eight, nine, ten, 11 steps, it can
Each key point wave band under the corresponding different-thickness of different materials workpiece is extracted, the feature vector of mutation waveform is calculated;
Step 13: setting superposition coefficient η, then obtain the wave that different wave mutation addition of waveforms caused by synchronization generates
The feature vector of section;
Step 14: self study typing new feature vector, can be arranged self study mode with support vector machines;
Step 15: amendment edge data;
Step 16: establishing database, all of above feature vector is imported into database according to the classification of type of mutation waveform.
2. a kind of method for establishing wire cutting power curve mutation waveform characterization libraries according to claim 1, feature exist
In:
Feature vector in the first step are as follows: mean valueWherein t1And t2Respectively thus wave band initial time,
The time is terminated, p indicates wire cutting power curve function, amplitudeWherein pmaxWith pminDistinguish wave band thus
Very big, minimum, two pole line slopes:Wherein tmaxAnd tminRespectively wire cutting power curve
The maximum p in this wave bandmaxWith minimum pminCorresponding time point;In the second step, calculates and obtain wave band
Three-dimensional feature vector is respectively mean value pjsyWith amplitude pmsyAnd two pole line slope kssy;In the third step, extract
Power waveform in feature vector be respectively as follows: mean value pjsyz, amplitude pmsyzAnd the slope k of two pole linessyz;It is described
The 4th step in, calculate mutation wave band feature vector be respectively mean value pjk, amplitude pmkAnd two pole line slope kk;Institute
In the 5th step stated, the feature vector for calculating this wave band is respectively mean value pjkz;Amplitude pmkz;Two pole line slope kskz;It is described
The 6th step in, the feature vector for calculating this wave band is respectively mean value pjrWith amplitude pmrWith two pole line kr;Described the 7th
In step, the feature vector for calculating this wave band is respectively as follows: mean value pjrh, amplitude pmrh, pole line distance krh;8th step
In, the feature vector for calculating this wave band is respectively as follows: mean value pjrc, amplitude pmrcAnd two pole line slope krc;Described the 9th
In step, the feature vector for calculating this wave band is respectively as follows: mean value pjrq, amplitude pmrqAnd two pole line slope krq;Described
In ten steps, the feature vector for calculating this wave band is respectively as follows: mean value pjrz, amplitude pmrzAnd two pole line slope krz;Described
In 11st step, the feature vector for calculating this wave band is respectively as follows: mean value pjh, amplitude pmhAnd extreme point line slope kh;It is described
The 12nd step in, the feature vector for calculating this wave band is respectively as follows: mean value pjc, amplitude pmc, extreme point line slope kc。
3. a kind of method for establishing wire cutting power curve mutation waveform characterization libraries according to claim 1, feature exist
In:
Mutation caused by addition of waveforms caused by waveform difference critical event synchronization in 13rd step occurs
Wave band feature vector are as follows: mean value η1(pj1+pj2);Amplitude is η2(pm1+pm2);Two extreme point line slopes are η3(k1+k2),
Middle η1、η2、η3Respectively two different key points are overlapped hourly values, amplitude, superposition coefficient corresponding to extreme point line slope,
pj1With pj2、pm1With pm2、k1With k2The mean value, amplitude and extreme point line that respectively two different key points correspond to wave band are oblique
Rate.
4. a kind of method for establishing wire cutting power curve mutation waveform characterization libraries according to claim 1, feature exist
In: self study typing new feature vector in the 14th step, with the method for support vector machines setting self study mode are as follows: use
The sample vector that front obtains calculates the feature vector of new waveform after the new power curve waveform of typing as sample for reference
Mean value pjx, amplitude pmx, extreme point line slope kx, feature vector and sample vector are subjected to preliminary comparison according to formula (1), if
When formula (1) is set up, then it is assumed that new wave band feature vector is classified as the corresponding wave band of this kind of key points, then by sampling feature vectors with
New feature vector with formula (2) be according to making fine correlation, if formula (2) is all set up, i.e., by this wave band be determined as such sample to
Amount,
In formula (1), ε is similarity given threshold;In formula (2), ε1For amplitude pjGiven threshold, ε2For mean value pmSetting threshold
Value, ε3For the given threshold of extreme point line slope K.
5. a kind of method for establishing wire cutting power curve mutation waveform characterization libraries according to claim 1, feature exist
In: the method for database is established in the 16th step are as follows: firstly, establishing workpiece material thickness table, which includes " workpiece material
Material ", " thickness of workpiece ", " material thickness ID " three parameters, wherein " material thickness ID " is by " workpiece material " and " thickness of workpiece "
It determines;Then, mutation type of waveform table is established, which includes " mutation type of waveform ID " and " material thickness ID " two parameters,
" material thickness ID " is consistent with " the material thickness ID " in workpiece material thickness table in the mutation type of waveform table;Finally,
Mutation wave character vector storage table is established, which includes " mutation Band Class ID ", " mutation wave band feature vector ", " acquisition
Three parameters of time ", the 12nd step treated mutation Waveform storage mutation wave character vector storage table " mutation wave band
In type ID ", the feature vector for the mutation waveform that the 13rd step is extracted is stored in " the mutation of mutation wave character vector storage table
In wave band feature vector ", " the mutation Band Class ID " of mutation wave character vector storage table is " prominent with mutation type of waveform table
Become Band Class ID " to be consistent;
The step of being mutated waveform using the data base querying are as follows: " workpiece material " and " thickness of workpiece " determines in workpiece material thickness table
Fixed unique " material thickness ID ";It utilizes " material thickness ID " can find in mutation type of waveform table under the material thickness to store
Mutation waveform, i.e., " mutation type of waveform ID ";Recycle " mutation type of waveform ID " in mutation wave character vector storage table
In find the eigenvector information met the requirements, this database stores workpiece information, including material, thickness simultaneously, and mutation wave
Shape information, including type of waveform, feature vector, acquisition time;
Structured Query Language database can be used in database in 14th step, also can be used
2 database of DATABASE.
6. a kind of method for establishing wire cutting power curve mutation waveform characterization libraries according to claim 2, feature exist
In: the bath is mutated wave band mean value pjsWith amplitude pms, wire transporting be mutated wave band mean value pjsyWith amplitude pmsy, bath wire transporting open
When silk cylinder turn to mutation wave band mean value pjsyzWith amplitude pmsyz, open air transportion silk cutting mutation wave band mean value pjkWith amplitude pmk, it is empty
Wire transporting cutting wire cylinder turns to mutation wave band mean value pjkzWith amplitude pmkz, incision steel plate be mutated wave band mean value pjrWith amplitude pmr, incision
Silk cylinder turns to mutation wave band mean value p afterwardsjrzWith amplitude pmrz, vertical duction be mutated wave band mean value pjrcWith amplitude pmrc, arc cutting
It is mutated wave band mean value pjrqWith amplitude pjrq, angle cutting mutation wave band mean value pjrzWith amplitude pmrz, to be changed to n millimeters of work piece cuts prominent
Become wave band mean value pjhWith amplitude pmh, change other materials mutation wave band mean value pjcWith amplitude pmc, superposition wave band mean value be η1(pj1+
pj2) with amplitude be η2(pm1+pm2), new feature vector mean value pjxWith amplitude pmx, threshold epsilon, ε1、ε2、ε3Unit be w;Bath
It is mutated wave band pole line slope ks, wire transporting be mutated wave band pole line slope ksy, bath wire transporting open when silk cylinder turn to mutation
Wave band pole line slope ksyz, open air transportion silk cutting mutation wave band pole line slope kk, air transportion silk cutting wire cylinder turn to it is prominent
Become two pole line slope k of wave bandkz, incision steel plate be mutated two pole line k of wave bandr, silk cylinder turns to mutation wave band pole after incision
Point line distance krz, vertical duction be mutated wave band pole line slope krc, arc cutting mutation two pole line slope of wave band
krq, angle cutting mutation wave band pole line slope krz, be changed to n millimeters of work piece cuts mutation wave band extreme point line slope ksh、
It is η that other materials, which is mutated wave band extreme point line slope,3(k1+k2), superposition coefficient η1、η2、η3It is dimensionless number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611232737.1A CN108247163B (en) | 2016-12-28 | 2016-12-28 | A method of establishing wire cutting power curve mutation waveform characterization libraries |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611232737.1A CN108247163B (en) | 2016-12-28 | 2016-12-28 | A method of establishing wire cutting power curve mutation waveform characterization libraries |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108247163A CN108247163A (en) | 2018-07-06 |
CN108247163B true CN108247163B (en) | 2019-05-31 |
Family
ID=62719896
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611232737.1A Expired - Fee Related CN108247163B (en) | 2016-12-28 | 2016-12-28 | A method of establishing wire cutting power curve mutation waveform characterization libraries |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108247163B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN2257507Y (en) * | 1996-07-05 | 1997-07-09 | 中国科学院电工研究所 | Linear cutting machine processing source with detecting and control function |
CN101254560A (en) * | 2008-04-08 | 2008-09-03 | 哈尔滨工业大学 | Numerical control wire-electrode cutting impulsing power source with controllable current waveform |
CN101722342A (en) * | 2008-10-21 | 2010-06-09 | 阿奇公司 | Method and apparatus for controlling an electric discharge machining process |
CN102699453A (en) * | 2012-06-20 | 2012-10-03 | 无锡微研有限公司 | Electric spark machining method and device for micro inverted taper hole |
CN105397212A (en) * | 2008-08-29 | 2016-03-16 | 通用电气公司 | Electro Discharge Machining Apparatus and Method |
KR20160126760A (en) * | 2015-04-24 | 2016-11-02 | 원장연 | An elecrode setting method of CNC electric discharge machine |
-
2016
- 2016-12-28 CN CN201611232737.1A patent/CN108247163B/en not_active Expired - Fee Related
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN2257507Y (en) * | 1996-07-05 | 1997-07-09 | 中国科学院电工研究所 | Linear cutting machine processing source with detecting and control function |
CN101254560A (en) * | 2008-04-08 | 2008-09-03 | 哈尔滨工业大学 | Numerical control wire-electrode cutting impulsing power source with controllable current waveform |
CN105397212A (en) * | 2008-08-29 | 2016-03-16 | 通用电气公司 | Electro Discharge Machining Apparatus and Method |
CN101722342A (en) * | 2008-10-21 | 2010-06-09 | 阿奇公司 | Method and apparatus for controlling an electric discharge machining process |
CN102699453A (en) * | 2012-06-20 | 2012-10-03 | 无锡微研有限公司 | Electric spark machining method and device for micro inverted taper hole |
KR20160126760A (en) * | 2015-04-24 | 2016-11-02 | 원장연 | An elecrode setting method of CNC electric discharge machine |
Also Published As
Publication number | Publication date |
---|---|
CN108247163A (en) | 2018-07-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhou et al. | A clustering-Based KNN improved algorithm CLKNN for text classification | |
Lu et al. | Global study of beyond-mean-field correlation energies in covariant energy density functional theory using a collective Hamiltonian method | |
CN105930860B (en) | simulation analysis method for classification optimization model of temperature sensing big data in intelligent building | |
Zhou et al. | Text categorization based on clustering feature selection | |
Gao et al. | Improved information gain-based feature selection for text categorization | |
Lin et al. | A decomposition based multiobjective evolutionary algorithm with classification | |
CN106513879B (en) | A kind of spark discharge state recognition and detection method based on chaology | |
CN104573033B (en) | A kind of dynamic url filtering method and device | |
CN108247163B (en) | A method of establishing wire cutting power curve mutation waveform characterization libraries | |
Zhang et al. | Intelligent pulse analysis of high-speed electrical discharge machining using different RNNs | |
CN105583479A (en) | Electrolytic and mechanical combined machining servo control method and system based on short circuit rate | |
CN103984017A (en) | Automatic microearthquake focus positioning method | |
TWI700492B (en) | Molding characteristic mass spectrum and identification model establishment method and method of analysis and identification of microbial characterization | |
CN103995497B (en) | The complex-curved recognition methods of aircraft structure extended based on interarea | |
CN108255132A (en) | A kind of method for waveform identification based on wire cutting power curve mutation waveform database | |
Wankhade et al. | A survey: Approaches for handling evolving data streams | |
CN113466615B (en) | Drosophila optimization algorithm-based post-fault wave recording data synchronization method | |
CN102906653A (en) | Numeric control program generation method, device for same, and program for causing computer to execute method for same | |
Zhang et al. | Clustering centroid selection using a K-means and rapid density peak search fusion algorithm | |
CN110377798B (en) | Outlier detection method based on angle entropy | |
Li et al. | Out-of-step protection in generator based on intelligent identification of impedance trajectory | |
LI et al. | Dynamic particle swarm optimization algorithm based on automatic fast density peak clustering | |
Chen et al. | The influence of parameters of linear flat bottom chip breaker on chip flow | |
Gao et al. | Research on surface roughness of superalloy thin-walled parts with electrical discharge grinding based on orthogonal test | |
Yan et al. | Method for Detecting Anomaly Data of WAMS System Based on GA-iForest |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190531 Termination date: 20201228 |
|
CF01 | Termination of patent right due to non-payment of annual fee |