CN1430721A - Method and system of in-line monitoring punching procedure - Google Patents

Method and system of in-line monitoring punching procedure Download PDF

Info

Publication number
CN1430721A
CN1430721A CN01810063A CN01810063A CN1430721A CN 1430721 A CN1430721 A CN 1430721A CN 01810063 A CN01810063 A CN 01810063A CN 01810063 A CN01810063 A CN 01810063A CN 1430721 A CN1430721 A CN 1430721A
Authority
CN
China
Prior art keywords
strain
diagnosis
line monitoring
signal
carried out
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.)
Granted
Application number
CN01810063A
Other languages
Chinese (zh)
Other versions
CN1209597C (en
Inventor
徐扬生
杜如虚
唐珩
张桂才
葛明
梁润怡
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chinese University of Hong Kong CUHK
Original Assignee
Chinese University of Hong Kong CUHK
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chinese University of Hong Kong CUHK filed Critical Chinese University of Hong Kong CUHK
Publication of CN1430721A publication Critical patent/CN1430721A/en
Application granted granted Critical
Publication of CN1209597C publication Critical patent/CN1209597C/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
    • G05B19/406—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by monitoring or safety
    • G05B19/4065—Monitoring tool breakage, life or condition
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B30—PRESSES
    • B30B—PRESSES IN GENERAL
    • B30B15/00—Details of, or accessories for, presses; Auxiliary measures in connection with pressing
    • B30B15/26—Programme control arrangements
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/37—Measurements
    • G05B2219/37252—Life of tool, service life, decay, wear estimation
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/45—Nc applications
    • G05B2219/45137—Punch, stamp, also with use die, mould

Abstract

A method for in-line monitor of the punching procedure of one or more punch machines includes such steps as sampling multiple reference strain signals in normal punching information, smoothing them, determining reference curve, defining strain tolerance range, acquiring the in-line strain signals of punch machines by strain sensor, judging if the signals are in said strain tolerance range, and controlling the punch machines. Its system is composed of the units which can perform said all steps one for one.

Description

The method and system of on-line monitoring punching course
Technical field
This patent has not only been developed a cover to one or many on-line monitorings of punch press in ordinary production and the complete method of diagnosis, and has set forth and how to have utilized advanced small echo method for processing signals to construct effective monitoring algorithm to realize that the branch folding is monitored and diagnosed to one or more punch press simultaneously.
Background technology
Metal punching course is a kind of mechanical processing process commonly used, it simultaneously also is a very complicated process, it makes metal produce very big elastic deformation and plastic yield tens of in hundreds of milliseconds time, metal blank becomes the different shape that people need the most at last, need to consume lot of energy in this transient process, the related process variable of punching course is above 40.Domestic and international for many years many scholars have carried out a large amount of research to the monitoring of punching course with control from different angles, comprising " Jin and Shi used wavelet transformto compress the tonnage signal into a small set of attributes and then usedan Artificial Neural Network (ANN) for classification; (Technometrics; November 1999; Vol.41; No.4 pp.327-339.) " or the like. however, their majorities all belong to theoretical research, are difficult in the actual production environment.
Summary of the invention
This patented invention is intended to carry out on-line monitoring and diagnosis provides a cover complete method and operating system for the production operation of or many punch presses.Its principle is that the piezoelectric strain sensor is installed on the column of punch press, so that extract the strain signal in the punching press letter news under the ordinary production situation, and try to achieve the low frequency principal component of strain signal (indirectly stamping press) with the method for wavelet decomposition and reconstruct, obtain the mean value and the variance of this principal component according to the sample signal of several normal conditions, thereby can obtain the last lower curve alarm limit (just " tracking threshold value ") of normal condition by self-teaching, when monitoring, calculate the strain signal low frequency principal component of each stroke earlier, compare with the alarming line that configures then, any point transfinites and then sends alerting signal and punch press is stopped automatically, in order to avoid cause more waste products.Mould and punch press are further protected, improved the quality of products, the purpose of reduction in the numbers of seconds to reach.
This patented invention is intended to the production operation of or many punch presses is carried out on-line monitoring and diagnosis provides a cover complete method, comprises the following step:
A) gather a plurality of in the punching press letter news of punch press lathe bed under the ordinary production situation with reference to the strain signal sample;
B) by smoothly with reference to strain signal, determine reference curve;
C), set the strain tolerance scope of surveying by the punch press lathe bed according to reference curve;
D), gather the online strain signal of punch press lathe bed that records by selected strain transducer;
E) whether the online strain signal of diagnosis drops in the margin tolerance; And
(f) according to the result who in step e), draws, the control machine operation.
This patented invention also is intended to the production operation of or many punch presses is carried out on-line monitoring and diagnosis provides an operating system, comprises:
One is installed in the device on the punch press lathe bed, in order to obtain the punching press strain signal;
Through strain transducer, a plurality of parts on the punch press under the collection ordinary production situation with reference to strain signal;
Processing is with reference to strain signal, to determine the parts of reference curve;
According to reference curve, set the parts of the strain tolerance scope of surveying by the punch press lathe bed;
By selected strain transducer, gather the parts of the online strain signal of punch press lathe bed that records;
Diagnose online strain signal whether to drop on the interior parts of margin tolerance; And
The parts of control machine operation.
Relevant detailed content of the present invention and technology are as follows with regard to accompanying drawings now:
Description of drawings
Fig. 1 has described working condition and the environment that " C " type framework punch press is included in line monitoring system.
Fig. 2 cording volume description the structure of " C " type framework punch press mould in punching course among Fig. 1.
Fig. 3 ties up to basic structure diagram (comprising internal hardware and external communication), workflow diagram and the data transmission figure of line supervisory system.
Fig. 4 ties up to the flow chart of the concrete work of line supervisory system.
Fig. 5 ties up to the line supervisory system is carried out signal extraction work when normal punching press idiographic flow table.
Fig. 6 describes free-running frequency numerical value and the selected thus sensor installation site figure that modal analysis test is determined.
Fig. 7 system is described in the punching course strain curve that obtains from the piezoelectric strain sensor five zoness of different a press cycle.
Fig. 8 is typical strain curve figure under the one-shot situation, and it comprises (a) normal condition, (b) workpiece material thickening, and c) waste material is beated.
A-G among Fig. 9 is typical strain curve figure under the continuous punching press situation, and it comprises normal condition, workpiece material misdelivery and occur in the situation that the waste material of diverse location is beated.
A-B among Figure 10 is under the one-shot situation and the comparison of the strain curve figure under the continuous punching press situation.
Symbol description among the figure
10---wiring layout 36---charge amplifier 60---in-service monitoring
12---in-service monitoring device 38---display screen 62---LCD shows
14-16---punch press 40---punch press control 64---user is interrupted
18---CD-ROM drive motor and the current statistical report form of flywheel 42---alarm lamp 66---
The 20---patrix 44---68---of CCC data transmission
The last punching pin 46---of 22---processor 80---number of samples
24---counterdie 48---user profile 84---statistical calculation
Punching pin 50---sampling parameter 88---" tracking threshold value " under the 26---
Bound
28---piezoelectric strain sensor 52---warning is provided with the 92---drift and is parked on the dead-centre position
30---3 dimensional force sensor 54---self study is reported to the police and is provided with
32---laser sensor 56---imports sample length
34---data acquisition system (DAS) 58---on-line sampling and self study thresholding are provided with
Embodiment
The definition of " tracking threshold value " if not otherwise specified, so (or follow the tracks of and report to the police) is equal to the definition of " margin tolerance ".
Because punching course is the transient process of a duration very short (most of punching course is tens of approximately so that hundreds of milliseconds from the drift contact workpiece to the time of leaving workpiece).And also be typical transient signal by the strain signal that punch press lathe bed column is surveyed, it can in a press cycle, resolve into different zones and with this circulation in different time corresponding.As shown in Figure 7, on regional 1 section, strain signal approaches null value, and drift is in the preliminary work shape.On 2 sections in the Zai District territory, the holding pad contact workpiece makes workpiece produce slight deformation and also correspondingly makes strain value increase by a certain percentage simultaneously.On 3 sections, the drift contact workpiece also applies stamping press to workpiece in the zone, and workpiece deformation should increase strain facies.On 4 sections, stamping press surmounts in plastic deformation of metal material power in the zone, and plastic deformation of metal mainly betides in this process, and stamping press moment sharply descend, strain signal has the process of a decay concussion.On 5 sections, drift begins to turn back to the initial position of preliminary work in the zone, and a press cycle finishes.Strain signal comprises the data information of static and dynamic perfromance thus, and it can be applicable to online monitoring system effectively.
The signal characteristic of piezoelectric strain sensor can be known the abnomal condition of workpiece in actual production process effectively.In fact, when thickness of workpiece increases, the rising edge of signal trends towards towards left to moving, as shown in 2,3 sections in the zone among Fig. 7.This displacement now can shift to an earlier date contact workpiece as just having reacted drift.If the situation that has waste material to beat takes place, signal just trends towards producing very big strain value, and this existing picture has just reacted unbalanced stamping press and caused strain value to become big.All these features make the column strain signal can fully reflect the characteristics of metal stamping on-line monitoring.
Certainly, different fault signatures may occur in the different part of strain signal.Relatively more difficult in addition is to extract signal from the punch press lathe bed to monitor the often too much noise of introducing signal of punching press situation, thereby is difficult to obtain level and smooth strain signal.
As everyone knows, Fourier conversion (or progression) is with signal transformation or decomposes the sine wave of different frequency, amplitude and phase place, and sine wave is a function of overall importance, local feature that can not reflected signal.Decompose different with Fourier, wavelet decomposition then be with signal decomposition to a series of tight support functions with different concussion frequencies, these a series of finite support and the function with concussion characteristic then are to be obtained by expansion or compression and translation by the same function (or abbreviating wavelet function as) that is called as female small echo.The continuous wavelet transform of an arbitrary signal x (t) is defined as: W ψ x ( a , b ) = 1 a ∫ - ∞ ∞ x ( t ) ψ * ( t - b a ) dt
Wherein: Ψ (t) is (mother) wavelet function, and a (a＞0) is an expansion factor (or being called scale factor), and (shift factor of b＜R), ' * ' represents complex conjugate operation to b.
Continuous wavelet transform is because calculated amount is big and do not have a fast algorithm, initial stage in the wavelet theory development has been limited its application greatly, Mallat with scale factor a and shift factor b carry out two advance to disperse and introduce fast algorithm after, the application of small echo has just obtained great development.With b and a carry out two advance discrete, that is: a j=2 jAnd b J, k=k/2 j, can define orthogonal wavelet transformation.If family of functions
Ψ M, k(t)=2 M/2Ψ (2 mT-k), m, k ∈ Z constitutes an orthogonal basis, and the orthogonal wavelet transformation that then obtains is: x k m = ∫ - ∞ ∞ x ( t ) ψ m , k ( t ) dt
The orthogonal wavelet transformation result of signal is to be the frequency range that a series of bandwidth do not wait with signal decomposition, and the low-frequency range bandwidth is little, and the bandwidth of high band is big, and bandwidth increases progressively with scale-of-two, and figure below is that dyadic wavelet decomposable process frequency range is divided synoptic diagram.Utilize this characteristic can carry out the filtering of various passbands (single passband or many passbands) to signal.The monitoring algorithm that this patent proposes utilizes the filtering property of small echo to finish, and obtains smooth low frequency punching press strain characteristics curve by the decomposition to signal with the reconstruct of choosing band.
Figure A0181006300101
Dyadic wavelet decomposable process frequency range is divided synoptic diagram
This patent proposes the method for a kind of what is called " tracking threshold value " (or follow the tracks of and report to the police), can replace segmentation, no matter variation constantly takes place in the variation or each key point that are the size of stamping press, this curve threshold value can capture the change of the punching press state that different faults causes exactly, test findings shows that this method is very effective.This method adopts the method for wavelet decomposition and reconstruct, try to achieve the low frequency principal component of strain signal (stamping press indirectly) earlier, obtain the average and the variance of this principal component according to the sample signal of several normal conditions, thereby can obtain the alarm limit up and down of normal condition, when monitoring, calculate the strain signal low frequency principal component of each stroke earlier, compare with alarming line then, if any point transfinites and then sends alerting signal.
Here the monitoring algorithm of Ti Chuing is based on probability and mathematical statistics theory, need to gather earlier the stamping press signal M group (can choose different groups according to specific circumstances and count M=30～50) of normal punching press state before calculating, suppose that every group of signal has N data point, adopt discrete wavelet that every group of signal carried out five level decomposition, characteristics according to the punching press strain signal, only adopt the wavelet coefficient of a frequency range of lowest frequency to be reconstructed, obtain the level and smooth punching press strain signal after the de-noising, we obtain the level and smooth stamping press signal of M group like this, then to this its average of M group calculated signals: X ( i ) = 1 M Σ j = 1 M x i ( j ) , i = 1,2 , Λ , N And variance: S 2 ( i ) = 1 M - 1 Σ j = 1 M [ x i ( j ) - X ( i ) ] 2 , i = 1,2 , Λ , N Consideration on the counting yield, can adopt following equivalent formula: S 2 ( i ) = 1 M - 1 [ Σ j = 1 M x i 2 ( j ) - 1 M ( Σ j = 1 M x i ( j ) ) 2 ] , i = 1,2 , Λ , N
We can obtain two smooth last lower threshold values limits like this: X (i) ± CS (i), and=1,2, Λ, N according to the mathematical statistics theory, during C=3, will have 99.7% punching press strain smooth curve in this threshold value limit during normal punching press state.
After the stamping press threshold value limit that obtains two normal punching press states, the punching press strain signal that later each press cycle is recorded all carries out same wavelet filtering like this, makes comparisons with these two threshold value limits then, transfinites and promptly makes warning.
Process in this patent and system can be used for monitoring simultaneously one or more punch press.
In the system of patent, one or more strain transducers are installed on the lathe bed of punch press, to obtain the punching press strain signal.The sensor that can be used for punching course monitoring has multiple, for example, and force transducer, proximity sensor, strain transducer, acceleration transducer.According to test findings, from easy for installation, economic easy-to-use and monitoring effect, the piezoelectric strain sensor is ideal.
If monitoring target is " C " type framework punch press list work step mould, can only select a sensor for use, if multitask step step-by-step movement mould then has at least two sensors need be mounted respectively on two root posts of punch press.
In the method for patent, obtain high reliability in order to make monitoring, need to gather a plurality of normal punching press strain signals and handle, thereby set Threshold value curve.Obviously, many more samples will bring high more accuracy.Usually, at least 30 samples are preferably arranged.
The reference strain signal that a) obtains in the step in the method for patent the obtains level and smooth reference curve through wavelet decomposition and reconstruct.According to the definition of this patent method, from b) go on foot e) step, can come complete storage, network and more processing with the counter of Linux C++ program with a dress.
For a single work step mould, the strain signal of measuring from the lathe bed of punch press is simple (Figure 10 .A) relatively, and most of punching press fault can detect with the peak value feature.Yet for the stepping mould in a multitask step, its strain signal is very complicated (Figure 10 B).Because many peak values are arranged in these signals, it is quite difficult that the method for depending merely on peak value detects.In this case, two road strain signals of gathering from the right and left lathe bed are used to reflect more complete process feature.
Strain passes sensor should be installed in the easiest position that measures strain signal on the punch press, in order to determine this position, we have carried out modal analysis test on " C " type framework punch press, overlap special-purpose model analysis software with 4 acceleration transducers and in the test, according to the data that test records, determine several fixed frequencies: 38Hz, 68Hz, 140Hz, 260Hz.Test findings finds, sensor sensitive installation site is positioned at the middle part (140Hz) of " C " type framework punch press, and this position is the weakest position of such punch press.The data of the model analysis of punch press are very important concerning punching press monitoring, because the position that they can aiding sensors be installed is determined in the most responsive position of pressure condition that liquidates.
The method of definite reference curve of this patent invention is as follows.The piezoelectric strain sensor is to make according to the piezoelectric crystal structure can discharge electric charge under action of mechanical pressure principle.One or more significant relevant with punching course in time signal parameters compare with determining good reference value in advance then via the strain transducer monitoring that is installed on the punch press.One is adopted the trigger of internal trigger signal to guarantee that the starting point of each sampling all is consistent, and generally the activation threshold value of this trigger is set at 0.2V.The collection of sample is that 2～5KHz carries out with internal trigger and sample frequency, and the sampling period has been covered punching press Huis circle.The setting of " tracking threshold value " usually relies on monitored punching press situation, and the marquis need not set up several groups " tracking threshold values " to monitor complex operating environment in monitoring respectively when cutting off major part.
Metal stamping on-line monitoring system of the present invention is intelligent, reliable, relative low price.This system mainly contains following characteristics:
1. combine the signal feature and the independently developed stable artificial intelligence control method of high performance good quality sensor; according to the characteristics of signals of mould punch press fuselage when bottom dead centre is worked change that the waste material that comes in the on-line monitoring punching course is automatically and accurately beated, common problem in the punching press such as bending, material thickness variation, misdelivery and two are beaten again; and control the punch press motion automatically, mould and punch press are further protected.Improve the quality of products thereby reach, reduction in the numbers of seconds gets purpose.
2. computerized flexible means; can be simultaneously to one or more punch press/mould motion carrying out monitoring; reduce user's cost greatly; and can write down and transmit user's data automatically with network; for example: add up automatically, the quantity of bad product; shut down number of times and time, situation analysis of mould/punch press and operator or the like.For producer designs at mould, machine operation is so that aspects such as production control provide the analysis data of usefulness.
3. no matter be " C " type framework punch press or planer-type punch press, can both under the situation of not overslaugh operating and setting and punch press/mould motion, this product be installed safely, greasy dirt, the material bits, external factors such as dust and bad working environment can not constitute influence.And show the motion conditions of real-time punch press/mould with Chinese clear and easy to see (numerous/simplified Chinese character) or English picture.
In conjunction with following accompanying drawing, can further describe this patent invention.
Fig. 1 is that the principle of typical " C " type framework punch press is shown wiring layout (10) thoroughly.In this icon, stamping line comprises a pair of punch press (14-16).In an operation cycle, each punch press all can carry out certain metal forming processing.For example, these punch presses can sequentially carry out the continuous stepping die stamping operation and the punching operation of one-shot.According to the data (as the piezoelectric strain sensing data of handling through amplifier) of wavelet decomposition technology and the punch press (14-16) that records, the production operation that in-service monitoring device (12) can be carried out each punch press monitors.
Tong Time, this in-service monitoring device also can the generative process control informations, are used to be presented at LCD display and upload CCC.In this invention system, supervisory programme can allow the user at an easy rate from the operation punch press on or CCC obtain data.The user interface of supervisory programme has adopted Linux C++ and network technology.Selectable user instruction through touch-screen or other input equipment, is finished by multilevel menu.For general purpose, carry out data analysis and diagrammatic representation, this supervisory programme has adopted many library files, and uses the Linux C Plus Plus to programme.
Supervisory programme can move on the counter of any standard that the multifunctional data acquisition card that is used for data acquisition has been installed, as IBM compatible or Pentium programmable calculator.This counter should have the RAM storer of 128MB at least, the color graphics card of a VGA compatibility and a LCD display.Because of supervisory programme need be stored lot of data, so the hard disk that should select minimum 1GB for use is as main storer.In addition, the information that uploads to CCC through communication network comprises the position of punch press, and relevant triggering counter sends the data of the situation of signal.If CCC is a telnet, all signals and production information will be by the line network inputs of punch press.
As shown in Figure 2, " C " type framework punch press comprises a patrix (20), goes up punching pin (22), a counterdie (24) and a following punching pin (26) for one.CD-ROM drive motor and flywheel (18) normally by a belt driver, clutch coupling, detent be used for the master driver crank punch press and that have crank-bar linkage with one, a pitman shaft is connected with a drift.Alternative process will be finished in maximum 1/4 times of a full stroke of drift.Punch press has a pull bar upwards to expand to crank-bar linkage from CD-ROM drive motor.It is used for mobile patrix and last punching pin on downward direction.Piezoelectric strain sensor (28) is installed in the most responsive position of pressure condition that liquidates in as shown in Figure 6 the model analysis.
As shown in Figure 3, counter is furnished with the processor (46) that input/output interface is provided to system.Through the delivery outlet of punch press control (40) and central authorities' controls (44), boundary member of processor (46) may command.Unusually take place when detecting any punching press, boundary member will be controlled the motion of drift, and make it be parked in upper dead center.Like this, processor can be controlled the motion of punch press through boundary member, and with alerting signal of alarm lamp (42) output, a failure message is arranged simultaneously to display screen (38).Because this control has greatly reduced the production fault, so punch press and mould have obtained protection.Two piezoelectric strain sensors (28) are connected on the punch press through a data acquisition system (DAS) (34) and a charge amplifier (36).Can transmit the input port of arriving counter processor (46) through the data of software generation like this.According to the type of mould, polytype 3 dimensional force sensors (30) and laser sensor (32) can be used for system monitoring.CCC (44) adapts to the signal that reception is transmitted through communication network from processor (46), and shows the supervision situation of punch press.
Especially, with reference to figure 4, it has shown that this is used for the workflow of the invention on-line system of supervision and diagnosis " C " type framework punch press.As shown in the figure, this system can monitor and diagnose many features, as " user profile " (48); " sampling parameter " (50); " report to the police and be provided with " (52), " self study is reported to the police and is provided with " (54), " input sample length " (56); " on-line sampling and self study thresholding are provided with " (58); " 24 hours and 7 days in-service monitorings " (60), " LCD demonstration " (62), " user's interruption " (64); " current statistical report form " (66) and " data transmission " (68).Under the control of counter, punching operation is by with (approximately 2KHz～5KHz) the sampling repeatedly of certain sample frequency.In order to obtain accurate sample waveform, counter is according to the different sampling process control internal trigger levels and the setting of gain.Stored samples size (56) is represented the punching press waveform with the amplitude of time and stamping press, and is transmitted to processor.With an equation of its substitution (58) after a warning thresholding is automatically learned out in statistical computation.Whether the peak value of the signal that inline diagnosis shown in Figure 4 can be obtained according to two piezoelectricity modification sensors (28) and energy value drop on is diagnosed stamping system adaptively by " tracking threshold value " determined allowed band of optimum in advance.If any one the upper limit (or less than lower limit) in these two eigenwerts greater than respective range, in conjunction with alarm lamp (42), punch press control system (40) with the punch press stop at upper dead center (TDP).This arrangement helps adjusting punch press, and the burden that reduces operating punch machine person also greatly reduces the rejection rate of product.For display position or debate the number of not unusual mould, the user is interrupted (64) available LCD display (62).This recommended arrangement easy operating person offers an explanation unusual mould or punching press fault, and helps checking, adjusts and mold exchange more, improves the work efficiency of stamping system.
Figure 5 shows that the part process flow diagram of the metal stamping monitoring system of this patent invention.In native system was realized, the punching press waveform was repeated sampling with certain sample frequency under computer control, and the number of samples 80 of normal punching press state should be more than 30 at least.The size of sample frequency can correspondingly be changed by computer control according to the speed of punching press, the waveform sampling data that can guarantee different punch process is like this counted identical, the data 82 that sampling obtains is carried out statistical calculation 84 by computing machine, can be obtained the bound 88 of " tracking threshold value " by the average and the standard variance of each material point.In the on-line monitoring of process, to the punching press waveform of each sampling, if any sampled point exceeds the threshold value bound that configures, the alarm lamp on the monitoring system is bright, and controls 40 drifts and be parked on the dead-centre position 92.
Figure 6 shows that the mode of " C " type framework punch press, as can be seen from the figure, sensitive part should be in the middle and lower part of punch press, and strain transducer 28 should be installed in this position.
By shown in Figure 7, the punching press strain signal can be divided into the several characteristic section.Among the section S1, strain approaches zero, because punching course does not begin as yet; Among the section S2,, the binder version cause on the column strain to begin to increase because beginning contact workpiece; Among the section S3, the mould contact workpiece also applies stamping press to workpiece and makes workpiece begin to produce elastic-plastic deformation, and therefore in this section, strain value increases considerably; Among the section S4, stamping press increases makes workpiece material surpass yield point, and workpiece begins to break away from material, causes strain to descend rapidly; At last, leave workpiece at section S5 mould, the power that acts on the lathe bed discharges, and lathe bed begins the position of recovering original, and strain signal has fuctuation within a narrow range, and is returned to null value gradually.Examine and can find, strain signal previous section and stamping press variation are similar, and (S1～S4), the part of the back then variation tendency with vibration is similar.
Fig. 8 shows the strain signal of three kinds of different conditions of a kind of one-shot process.(a) normal 104, (b) workpiece is too thick 106, and (c) waste material beats 108.As seen from the figure, the strain signal pressure condition sensitivity that liquidates, when workpiece was too thick, a section S2 and a section S3 moved to left, and this is because the too thick result who causes mould to do sth. in advance contact workpiece of workpiece.The feature that waste material is beated then shows as strain signal by a relatively large margin fluctuation when punching course finishes.
Punching press monitoring system of the present invention has some advantages that are different from traditional supervising device.As shown in Figure 9, threshold curve is a kind ofly to set " tracking threshold value " by wavelet analysis technology.The algorithm of threshold setting can be described below:
1. measure 20～30 groups of sample signals of normal condition, every group of signal is for surveying certainly in two strain signals of left and right pillar.
2. use wavelet decomposition and reconfiguration technique, we obtain the low frequency component of filtered every group of signal, and this low frequency component signal is much more level and smooth than original signal, can be advantageously used in the setting of threshold curve.
3. calculate the average curve and standard deviation curve: the m of every group of filtering signal l(t) expression left column strain mean value signal, m r(t) expression right column strain mean value signal, v l(t) expression left column strain standard deviation curve, v r(t) expression right column strain standard deviation curve.
Can obtain two threshold curve: m up and down of left and right pillar strain signal thus respectively l(t)+3v l(t) and m l(t)-3v l(t) 110; And m r(t)+3v r(t) and m r(t)-3v r(t) 112, see Fig. 9-A.
Fig. 9-B is depicted as the tracking threshold curve 114,116 that utilizes wavelet analysis technology to set.
, utilize by wavelet analysis technology preset threshold curve to Fig. 9-G with reference to figure 9-C, the various most common failures of punching course beat 122 as changes in material, misdelivery 118,120, waste material continuously, 124,126,128,130,132,134,136, two beat again or the like, can be detected.In general, different punching press faults can reveal different features in the different sector table of strain signal, yet, accurate segmentation is very difficult thing according to different punching courses, the method that adopts us to propose, monitoring system can be easy to detect the process status variation that betides different sections in the punch process process continuously.
For single work step mould, the strain signal that is recorded by the punch press column is simple relatively, and (Figure 10-A) 138, and many faults can be come out by the change-detection of peak value stamping press.Yet, for multitask progressive die tool step by step, the strain signal that records by the punch press column with regard to complexity how (Figure 10-B) 140,142.In this case, in order to detect more punching press fault, the strain signal that measures bilateral column is very necessary.Different faults may betide the different sections on different strain signals (left side or right).

Claims (15)

1. method that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses comprises the following step:
A) gather a plurality of in the punching press letter news of punch press lathe bed under the ordinary production situation with reference to the strain signal sample;
B) by smoothly with reference to strain signal, determine reference curve;
C), set the strain tolerance scope of surveying by the punch press lathe bed according to reference curve;
D), gather the online strain signal of punch press lathe bed that records by selected strain transducer;
E) whether the online strain signal of diagnosis drops in the margin tolerance; And
F) according to the result who in step e), draws, the control machine operation.
2. the method that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 1 wherein should be obtained from a strain transducer that is installed on the punch press lathe bed with reference to strain signal.
3. the method that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 2, wherein this strain transducer is the piezoelectric strain sensor.
4. as claim 1 or the 2 or 3 described methods that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses, wherein this collection is a plurality of handles with reference to strain signal, preferably measures at least 20 samples.
5. the method that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 1 wherein adopts wavelet decomposition and reconstruct to gather this level and smooth with reference to strain signal described in the step a).
6. operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses comprises:
One is installed in the device on the punch press lathe bed, in order to obtain the punching press strain signal;
Through described device, a plurality of parts on the punch press under the collection ordinary production situation with reference to strain signal;
Handle described with reference to strain signal, to determine the parts of reference curve;
According to described reference curve, set the parts of the strain tolerance scope of surveying by the punch press lathe bed;
By described strain transducer, gather the parts of the online strain signal of punch press lathe bed that records;
Diagnose described online strain signal whether to drop on the interior parts of described margin tolerance; And
The parts of control machine operation.
7. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 6, wherein this device comprises one or more strain transducers.
8. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 7, wherein this device comprises a signal amplifier.
9. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 7, wherein this strain transducer is the piezoelectric strain sensor.
10. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 8, wherein this signal amplifier is a charge amplifier.
11. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 6 wherein adopts wavelet decomposition and reconstruct to carry out described processing with reference to strain signal, determines reference curve.
12. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 6, wherein this punch press is " C " type framework punch press, comprises two root posts.
13. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 12, wherein this framework has at least two strain transducers to be mounted respectively on two root posts of punch press.
14. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 13, wherein this strain transducer all is the piezoelectric strain sensor.
15. the operating system that on-line monitoring and diagnosis are carried out in the production operation of or many punch presses as claimed in claim 7, wherein this strain transducer is installed on the most responsive position of the pressure condition that liquidates in the model analysis.
CNB018100635A 2001-12-14 2001-12-14 Method and system of in-line monitoring punching procedure Expired - Fee Related CN1209597C (en)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2001/001630 WO2003052533A1 (en) 2001-12-14 2001-12-14 Method and system for on-line monitoring stamping operation

Publications (2)

Publication Number Publication Date
CN1430721A true CN1430721A (en) 2003-07-16
CN1209597C CN1209597C (en) 2005-07-06

Family

ID=4574905

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB018100635A Expired - Fee Related CN1209597C (en) 2001-12-14 2001-12-14 Method and system of in-line monitoring punching procedure

Country Status (5)

Country Link
CN (1) CN1209597C (en)
AU (1) AU2002218957A1 (en)
HK (1) HK1053866A1 (en)
TW (1) TW515738B (en)
WO (1) WO2003052533A1 (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101322084B (en) * 2005-12-05 2010-08-11 西门子公司 Method and control device for determination of the time duration before a machine element requires servicing
CN102489546A (en) * 2011-12-05 2012-06-13 湖北汽车工业学院 Method and system for measuring plastic deformation load
CN102520672A (en) * 2011-12-05 2012-06-27 湖北汽车工业学院 Method and system for monitoring plastic deformation process and defects
CN102741765A (en) * 2009-11-18 2012-10-17 株式会社日立制作所 System and method for extracting process signals
CN106180510A (en) * 2016-07-15 2016-12-07 洪燎 Multi-station cold-heading former control method and system
CN106990756A (en) * 2017-03-29 2017-07-28 大连理工大学 A kind of Digit Control Machine Tool geometric accuracy on-line monitoring method
CN109654991A (en) * 2018-12-29 2019-04-19 北京航天测控技术有限公司 A kind of dynamic strain measurement and analysis method based on edge calculations
CN111024527A (en) * 2019-12-06 2020-04-17 西安理工大学 Crack propagation monitoring method based on multi-sensor data fusion
CN113092149A (en) * 2021-03-31 2021-07-09 连杰 Intelligent punching press machine fault monitoring system, method and terminal based on MDDP system
CN113492162A (en) * 2020-03-18 2021-10-12 株式会社理光 Diagnostic device, diagnostic method, storage medium, and computer device
CN114433656A (en) * 2020-10-30 2022-05-06 深圳富桂精密工业有限公司 Punching abnormity detection system
CN114834085A (en) * 2022-06-29 2022-08-02 江苏双赢锻压机床有限公司 Centralized digital control method and system for online operation of multiple punching machines

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7216519B1 (en) * 2003-07-28 2007-05-15 Oes, Inc. Strain monitoring for part quality analysis
US9104650B2 (en) 2005-07-11 2015-08-11 Brooks Automation, Inc. Intelligent condition monitoring and fault diagnostic system for preventative maintenance
EP1913506A4 (en) * 2005-07-11 2008-08-13 Brooks Automation Inc Intelligent condition monitoring and fault diagnostic system for predictive maintenance
TWI493305B (en) * 2013-12-27 2015-07-21 Metal Ind Res & Dev Ct Machining Curve Setting and Displaying Method of Graphic Interface of Servo Punch
EP3308944B1 (en) * 2016-10-17 2021-04-14 Gabler Thermoform GmbH & Co. KG Method and apparatus for thermoforming
EP3608743A1 (en) * 2018-08-07 2020-02-12 KH Automotive S.r.l. Arrangement and method for controlling the pressing of metal sheets
TWI714097B (en) * 2019-05-24 2020-12-21 國立臺北科技大學 A detection system for die mold and life management therein
CN115169423B (en) * 2022-09-08 2023-05-02 深圳市信润富联数字科技有限公司 Stamping signal processing method, device, equipment and readable storage medium

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4837656A (en) * 1987-02-27 1989-06-06 Barnes Austen Bernard Malfunction detector
US4813320A (en) * 1987-08-11 1989-03-21 Oberg Industries, Inc. Method and apparatus for detecting a sheet strip material misfeed condition

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101322084B (en) * 2005-12-05 2010-08-11 西门子公司 Method and control device for determination of the time duration before a machine element requires servicing
CN102741765A (en) * 2009-11-18 2012-10-17 株式会社日立制作所 System and method for extracting process signals
CN102741765B (en) * 2009-11-18 2015-04-08 株式会社日立制作所 System and method for extracting process signals
CN102489546A (en) * 2011-12-05 2012-06-13 湖北汽车工业学院 Method and system for measuring plastic deformation load
CN102520672A (en) * 2011-12-05 2012-06-27 湖北汽车工业学院 Method and system for monitoring plastic deformation process and defects
CN102520672B (en) * 2011-12-05 2014-04-09 湖北汽车工业学院 Method and system for monitoring plastic deformation process and defects
CN102489546B (en) * 2011-12-05 2014-06-18 湖北汽车工业学院 Method and system for measuring plastic deformation load
CN106180510A (en) * 2016-07-15 2016-12-07 洪燎 Multi-station cold-heading former control method and system
CN106990756A (en) * 2017-03-29 2017-07-28 大连理工大学 A kind of Digit Control Machine Tool geometric accuracy on-line monitoring method
CN106990756B (en) * 2017-03-29 2019-03-05 大连理工大学 A kind of numerically-controlled machine tool geometric accuracy on-line monitoring method
CN109654991A (en) * 2018-12-29 2019-04-19 北京航天测控技术有限公司 A kind of dynamic strain measurement and analysis method based on edge calculations
CN111024527A (en) * 2019-12-06 2020-04-17 西安理工大学 Crack propagation monitoring method based on multi-sensor data fusion
CN111024527B (en) * 2019-12-06 2022-11-18 西安理工大学 Crack propagation monitoring method based on multi-sensor data fusion
CN113492162A (en) * 2020-03-18 2021-10-12 株式会社理光 Diagnostic device, diagnostic method, storage medium, and computer device
CN114433656A (en) * 2020-10-30 2022-05-06 深圳富桂精密工业有限公司 Punching abnormity detection system
CN114433656B (en) * 2020-10-30 2024-04-23 深圳富桂精密工业有限公司 Stamping abnormality detection system
CN113092149A (en) * 2021-03-31 2021-07-09 连杰 Intelligent punching press machine fault monitoring system, method and terminal based on MDDP system
CN113092149B (en) * 2021-03-31 2023-01-17 连杰 Intelligent punching press machine fault monitoring system, method and terminal based on MDDP system
CN114834085A (en) * 2022-06-29 2022-08-02 江苏双赢锻压机床有限公司 Centralized digital control method and system for online operation of multiple punching machines

Also Published As

Publication number Publication date
AU2002218957A1 (en) 2003-06-30
TW515738B (en) 2003-01-01
CN1209597C (en) 2005-07-06
WO2003052533A1 (en) 2003-06-26
HK1053866A1 (en) 2003-11-07

Similar Documents

Publication Publication Date Title
CN1209597C (en) Method and system of in-line monitoring punching procedure
CN101509839A (en) Cluster industrial robot failure diagnosis method based on outlier excavation
CN1129084C (en) Manufacturing process change control apparatus and its manufacturing process change control method
CN102520672B (en) Method and system for monitoring plastic deformation process and defects
CN105353726A (en) Special-purpose controller of electronic press, and control method
CN104808585A (en) Machine tool health state rapid checking method
CN111060337A (en) Running equipment real-time fault diagnosis method based on expert system
Zhang et al. Deep convolutional neural network using transfer learning for fault diagnosis
CN108062514A (en) A kind of ink roller of offset printing machine method for diagnosing faults based on three-dimensional spectrum analysis
CN1494973A (en) Safety cut-off of power tool with rotary tool bit
CN1987697A (en) Fault diagnostic system and method for under industrial producing process small sample condition
CN106607461A (en) Hydraulic AGC fault diagnosis expert system for cold-rolling mill
CN110346130A (en) A kind of boring flutter detection method based on empirical mode decomposition and time-frequency multiple features
CN202596709U (en) Oil well fault intelligent diagnosing device
CN101182802A (en) Consecutive pressurization system failure prediction, diagnostic equipment, failure prediction and diagnostic method
CN114714145A (en) Method for enhancing, comparing, learning and monitoring tool wear state by using Gelam angular field
CN108760268A (en) A kind of Vertical Mill operation data both phase step fault diagnostic method based on comentropy
CN113344275A (en) Floating platform wave climbing online forecasting method based on LSTM model
CN205157178U (en) Button dynamometry curvometer
CN1904907A (en) High-speed image matching detecting system and method
CN105843891A (en) Incremental online characteristic extraction and analysis method and system
CN115329942A (en) Bolt fastening device and method based on artificial intelligence
CN103324138B (en) Tester and testing method for numerically-controlled machine tool virtual transmission chain
CN207866454U (en) One kind monitoring agricultural machinery failure system based on subharmonic vibration
CN106907995A (en) A kind of cutter automatic detection system

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20050706

Termination date: 20141214

EXPY Termination of patent right or utility model