CN105022371B - Threshold value waveform generating - Google Patents
Threshold value waveform generating Download PDFInfo
- Publication number
- CN105022371B CN105022371B CN201410829332.0A CN201410829332A CN105022371B CN 105022371 B CN105022371 B CN 105022371B CN 201410829332 A CN201410829332 A CN 201410829332A CN 105022371 B CN105022371 B CN 105022371B
- Authority
- CN
- China
- Prior art keywords
- waveform
- threshold value
- observation
- characteristic point
- point
- 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
- 238000000605 extraction Methods 0.000 claims abstract description 49
- 238000011156 evaluation Methods 0.000 claims abstract description 46
- 230000002159 abnormal effect Effects 0.000 claims description 28
- 239000000284 extract Substances 0.000 claims description 14
- 238000002203 pretreatment Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 8
- 230000008859 change Effects 0.000 description 37
- 238000000034 method Methods 0.000 description 35
- 238000004519 manufacturing process Methods 0.000 description 14
- 238000012545 processing Methods 0.000 description 13
- 238000007689 inspection Methods 0.000 description 11
- 238000005070 sampling Methods 0.000 description 11
- 238000001914 filtration Methods 0.000 description 9
- 230000002950 deficient Effects 0.000 description 6
- 238000005259 measurement Methods 0.000 description 6
- 230000000630 rising effect Effects 0.000 description 6
- 238000003860 storage Methods 0.000 description 6
- 230000009471 action Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 230000015654 memory Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000003754 machining Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000010972 statistical evaluation Methods 0.000 description 2
- 241000406668 Loxodonta cyclotis Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/28—Testing of electronic circuits, e.g. by signal tracer
- G01R31/317—Testing of digital circuits
- G01R31/3181—Functional testing
- G01R31/3183—Generation of test inputs, e.g. test vectors, patterns or sequences
- G01R31/318314—Tools, e.g. program interfaces, test suite, test bench, simulation hardware, test compiler, test program languages
-
- 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
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R29/00—Arrangements for measuring or indicating electric quantities not covered by groups G01R19/00 - G01R27/00
- G01R29/02—Measuring characteristics of individual pulses, e.g. deviation from pulse flatness, rise time or duration
- G01R29/027—Indicating that a pulse characteristic is either above or below a predetermined value or within or beyond a predetermined range of values
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/01—Subjecting similar articles in turn to test, e.g. "go/no-go" tests in mass production; Testing objects at points as they pass through a testing station
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/28—Testing of electronic circuits, e.g. by signal tracer
- G01R31/317—Testing of digital circuits
- G01R31/3181—Functional testing
- G01R31/3183—Generation of test inputs, e.g. test vectors, patterns or sequences
- G01R31/318307—Generation of test inputs, e.g. test vectors, patterns or sequences computer-aided, e.g. automatic test program generator [ATPG], program translations, test program debugging
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03K—PULSE TECHNIQUE
- H03K3/00—Circuits for generating electric pulses; Monostable, bistable or multistable circuits
- H03K3/02—Generators characterised by the type of circuit or by the means used for producing pulses
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Abstract
The present invention obtains a kind of threshold value waveform generating, obtain observing waveform feature will not because of the really average average waveform that lost observing the smooth effect that bring of equalization of waveform, so as to generate with the addition of the time-axis direction for observing waveform deviation will not appear as observation direction of principal axis deviation true deviation threshold value waveform.Including：From the feature extraction unit (13) of observation waveform extracting characteristic point；The relevant evaluation portion (14) evaluated the characteristic point extracted by feature extraction unit (13)；Distinguished point based is observed the statistical disposition portion (15) of the statistical disposition of the time-axis direction and observation direction of principal axis of waveform；And the threshold value waveform generating unit (16) of the threshold value waveform of the normality for judging observation waveform is generated based on carrying out the obtained value of statistical disposition on time-axis direction and observation direction of principal axis.
Description
Technical field
It is used for the threshold value waveform generating of the threshold value waveform for the normality for judging observation waveform the present invention relates to generation.
Background technology
In the past, in order to prevent bad products from launching to market, whether manufactured product is well checked.It is known to have
A kind of inspection method checked manufactured product, this method provide stimulus to manufactured product, pass through biography
The action of product when sensor (sensor) measurement provides stimulus, the survey of measurement result is have recorded by being collected in the form of sequential
Amount waveform is compared to judge whether product is good with the threshold value set in advance as benchmark or threshold value waveform.
Such as in the decision method using threshold value waveform, to collect whether the waveform of record departs from threshold value set in advance
The upper limit waveform or lower limit waveform of waveform are as determinating reference.Because the decision method using threshold value waveform only needs given threshold ripple
Shape, therefore can apply to multiple fields.Especially there is practical significance in the case where carrying out multi items variable production.
The generation method of threshold value waveform has been recorded in patent document 1 and patent document 2.Method disclosed in patent document 1 will
The average waveform calculated according to multiple normal waveforms and the added standard deviation of above-mentioned multiple normal waveforms are used as upper notch
Shape, and above-mentioned average waveform and above-mentioned standard deviation are subtracted each other and are used as lower limit waveform.On the other hand, disclosed in patent document 2
Method is not that the average or standard deviation of multiple normal waveforms is evaluated, but the space that will be surrounded by multiple normal waveforms
As good space should be determined as, so as to generate upper limit waveform and lower limit waveform.
Prior art literature
Patent document
Patent document 1：Japanese Patent Laid-Open 2002-341909 publications
Patent document 2：Japanese Patent Laid-Open 2004-239879 publications
The content of the invention
The technical problems to be solved by the invention
However, in the method that patent document 1 is proposed, what is obtained after accessed multiple normal waveforms are averaged is flat
Equal waveform not necessarily real average normal waveform.In order to obtain really average waveform, it is necessary to obtain a number of
Normal waveform is averaged, if the acquisition quantity of normal waveform is one to two, is inadequate.On the other hand, if obtaining one
The normal waveform of fixed number amount is averaged, then the equalization of normal waveform can cause being smoothed of wave character, may lose
For determining whether good wave character, cause average waveform and non-real average waveform.
In addition, in the method for patent document 1, it is worth part, i.e. rising edge portion jumpy when containing in normal waveform
Point or during trailing edge part, the standard deviation of leading edge portion or trailing edge part easily becomes unstable.Due to the sight of waveform
Examine according to the triggering (trigger) of beginning and start, therefore, although will can be observed waveform at the time of be considered as it is identical,
Reality check object in caused physical phenomenon at the time of it is not completely the same, can with slightly at the time of deviation.On if
Rise along part and offset forwards, backwards on time-axis direction, then deviation can become observation in result at the time of leading edge portion
Deviation, and show in standard deviation.However, standard deviation showed actually not be observation deviation, but when
The deviation at quarter.Therefore, although actually observation does not have too large deviation, due to producing deviation because of instants offset, so as to
Cause to generate excessive threshold value waveform.That is, the waveform offsets of time-axis direction are considered as the waveform offsets of observation direction of principal axis, right
It carries out statistical evaluation, it is reflected as offset, therefore, in the threshold value waveform generated, particularly in leading edge portion
Or trailing edge part, do not necessarily mean that the correct upper limit or lower limit.
In addition, the upper limit and lower limit of the threshold value waveform of the method generation proposed using patent document 2 are rational, but only
It is inadequate, it is necessary to obtain with a certain degree of error width and suitably offset more only to obtain a number of normal waveform
Individual normal waveform.However, this is not easy to.It is and appropriate in order to verify whether to obtain with a certain degree of error width
Multiple normal waveforms of skew, as a result still need statistical evaluation as the standard deviation in patent document 1.
The present invention is to complete in view of the foregoing, and its object is to obtain a kind of threshold value waveform generating, it is obtained
There must be smoothing effect caused by the equalization of observation waveform and not lose the really average of the feature of observation waveform and put down
Equal waveform, and generation will not appear as the deviation for the time-axis direction for observing waveform the true inclined of the deviation of observation direction of principal axis
The threshold value waveform that difference is taken into account.
Technical scheme used by solution technical problem
To solve the above problems, realizing goal of the invention, the invention reside in a kind of threshold value waveform generating, generation judges to see
Survey threshold value waveform used in the normality of waveform, it is characterised in that including：Feature extraction unit, this feature extraction unit is from observation
Characteristic point is extracted in waveform；Relevant evaluation portion, the relevant evaluation portion is to the characteristic point that is extracted by the feature extraction unit
Evaluated；Statistical disposition portion, the statistical disposition portion are observed time-axis direction and the sight of waveform based on the characteristic point
The statistical disposition of measured value direction of principal axis；And threshold value waveform generating unit, the threshold value waveform generating unit be based on the time-axis direction with
And the statistical disposition on the observation direction of principal axis generates the threshold value waveform.
Invention effect
According to the present invention, have the effect that：The feature that waveform can be obtained observing will not be because of the equalization band of observation waveform
The smooth effect that comes and the really average average waveform lost, so as to obtain with the addition of the time-axis direction of observation waveform
Deviation will not appear as the threshold value waveform of the true deviation of the deviation of observation direction of principal axis, by using with the addition of true deviation
Threshold value waveform, so as to rightly to judging whether product good when the obtained waveform of measurement compare with threshold value waveform
Compared with.
Brief description of the drawings
Fig. 1 is the figure of hardware (hardware) structure for the embodiment 1 for representing the threshold value waveform generating of the present invention.
Fig. 2 is the figure for representing to perform the structure of the program (program) of the generation of threshold value waveform.
Fig. 3 is to represent that waveform configuration has the figure of the point of large change.
Fig. 4 is the figure of the simplest type for the angle change for representing waveform.
Fig. 5 is the figure for representing the first kind needed for the evaluation of one group of angle change.
Fig. 6 is the figure for representing the Second Type needed for the evaluation of one group of angle change.
Fig. 7 is the figure for the relation for schematically showing the character pair point between different waveforms.
Fig. 8 (a), Fig. 8 (b), Fig. 8 (c), Fig. 8 (d) are the characteristic point and characteristic point for representing to be extracted by feature extraction unit
The figure of correlation.
Fig. 9 (a) is that the figure for the average waveform for representing embodiment 1, Fig. 9 (b) are the marks for the observation for representing embodiment 1
Figure, Fig. 9 (c) of quasi- deviation are the figures of the standard deviation for the time for representing embodiment 1.
Figure 10 (a), Figure 10 (b), Figure 10 (c) are to represent to add observation to the average waveform calculated in embodiment 1
Standard deviation and the standard deviation of time and the figure of threshold value waveform generated.
Figure 11 is the figure for schematically illustrating the evaluation method of average waveform in threshold value waveform generating unit.
Figure 12 is the figure for the feature point extraction for representing that the smoothness adjustment of embodiment 2 is involved.
Figure 13 is the standard for schematically showing the observation direction of principal axis that the point group based on the observation in flat section is carried out
The figure of the evaluation of deviation.
Figure 14 is the figure of the structure for the embodiment 3 for representing the threshold value waveform generating of the present invention.
Figure 15 (a), Figure 15 (b), Figure 15 (c) are the abnormal thresholds produced corresponding to section for schematically showing embodiment 3
It is worth the figure of the generating process of waveform.
Figure 16 is the figure of the generation method for the threshold value waveform for schematically showing patent document 1.
Figure 17 is the figure of the corresponding relation of multiple observation waveforms in the statistical disposition for schematically show patent document 1.
Figure 18 (a) is that figure, Figure 18 (b) of multiple observation waveforms in the statistical disposition for schematically show patent document 1 are to show
Figure, Figure 18 (c) of average waveform are the statistics for schematically showing patent document 1 in the statistical disposition of meaning property expression patent document 1
Figure, Figure 18 (d) of processing Plays deviation are the figures of threshold value waveform in the statistical disposition for schematically show patent document 1.
Embodiment
Below, the embodiment party of threshold value waveform generating involved in the present invention and method is described in detail based on accompanying drawing
Formula.In addition, the present invention is not limited by present embodiment.
Embodiment 1.
First, for ease of understanding embodiment 1,16~Figure 18 of reference picture to the method for the waveform comparison of patent document 1 and
Its problem points illustrates.
Figure 16 is the figure of the generation method for the threshold value waveform for schematically showing patent document 1.
<The generation of the threshold value waveform of patent document 1>
As following inspection, i.e.,：Stimulus is provided to the product for being known as non-defective unit, to sample (sampling) interval delta
The inspection that measures of action of product when t is to providing stimulus, to inspection proceed by detection trigger and preservation starts
Moment t1, sampling being started simultaneously at, the end to inspection also carries out detection trigger and preserves finish time tn, while stops sampling,
The required time of one circulation (cycle) is tn-t1, and waveform points are set to n points.Further, since the time needed for checking is every
All it is once fixed, therefore, is recorded to proceed by the collection of waveform with the triggering that starts of a circulation, remembered in waveform collection
Record and start to terminate as waveform collection record is also afterwards, at the time of waveform points reach n points.As described above, by
The product known for non-defective unit provides stimulus and carries out waveform collection record omits the normal waveform illustrated to obtain.Equally, it is right
The product for being known as defective products also provides stimulus and carries out waveform collection record, obtains the unusual waveforms for omitting diagram.
Obtain multiple normal waveforms and unusual waveforms as described above, and multiple normal waveforms for being got based on these and
Unusual waveforms carry out defined threshold waveform.For example, average waveform m obtained from the multiple normal waveforms got are averaged is upward
Side is offset to generate upper limit waveform h, is offset downwards to generate lower limit waveform l.Average waveform m time t is represented with m (t)
The value of interior observation direction of principal axis, similarly, the value of upper limit waveform h observation direction of principal axis are expressed as h (t), lower limit waveform l's
The value of observation direction of principal axis is expressed as l (t).Confirm that accessed unusual waveforms depart from sampling interval t1~tn to give birth to
Into threshold value waveform, most at last upper limit waveform h and lower limit waveform l are defined as the threshold value waveform as the upper limit and lower limit.Institute
The upper limit waveform h and lower limit waveform l of defined threshold value waveform waveform points are also n points, and the sampling interval is Δ t.
<The judgement of the measured waveform of patent document 1>
Stimulus is similarly provided to the product as detection object, with sampling interval t1~tn and sampling interval
Δ t measurements provide action during stimulus, and the value for obtaining the observation direction of principal axis in time t eliminates figure by what f (t) was represented
The measured waveform shown.Measured waveform whether depart from threshold value waveform judgement pass through the value of measured waveform each point and threshold value waveform is each
The value of point is compared to carry out.For example, upper limit waveform h and lower limit waveform l both sides are being set as shown in Figure 16 as threshold value
In the case of waveform, if can confirm that, the value of measured waveform each point is less than the value of upper limit waveform h each points, and each more than lower limit waveform l
The value of point, that is l (t1) ＜ f (t1) ＜ h (t1), l (t2) ＜ f (t2) ＜ h (t2) ..., l (tn) ＜ f (tn) ＜ h (tn), then sentence
It is set to measured waveform without departing from threshold value waveform, product is non-defective unit.
In the above example, transverse axis is set to time or sampling number, the longitudinal axis is set to observation, but thus will observe
In the case that waveform is transformed to the not waveform of commensurate, such as Fourier transformation (Fast Fourier Transform：FFT),
Transverse axis after then converting is changed into frequency, and the longitudinal axis is changed into the size of frequency component, is in other words frequency spectrum (spectrum).In Jiang Guan
In the case that survey waveform is converted to the not waveform of commensurate, if having preset the threshold value waveform corresponding to waveform after conversion,
It can similarly be judged.
Product after being processed as described above to manufacture is detected and judges the whether good patent document 1 of product
, it is necessary to set inspection operation after manufacturing procedure is manufactured in mode, each product is since manufacture to the time needed for completion
It can be equally long because there is provided inspection operation.In order to solve this problem, following method be present：By confirming manufacture processing
Waveform when whether waveform obtained from control instruction is measured in process or manufactures machining status departs from normal manufacture processing comes
Judge whether product is good.Whether confirm waveform obtained from control instruction or manufacture machining status are measured in manufacture manufacturing procedure
The method of waveform when departing from normal manufacture processing be referred to as checking on On line inspection (in-line inspection) or machine or
(on-machine verification) is verified in machine.
Check that the waveform decision method of use can also be identical with above-mentioned example on machine.That is, can not also be to manufactured
Product provides stimulus, measures the action of product when providing stimulus, and the manufacture for one product of measurement that replaces adds
The situation of the situation of a work, i.e. circulation, and by resulting waveform and upper limit waveform set in advance or lower limit waveform comparison
.
On the other hand, above-mentioned average waveform m is needed by rule of thumb to determine to the amount of downward shift.It is multiple that can get
In the case of normal waveform and multiple unusual waveforms, offset can be adjusted, it is good so as to which these normal waveforms are determined as, by this
Unusual waveforms are determined as bad a bit, but in order that the amount adjusted is suitable, it is necessary to obtain a number of normal waveform and exception
Waveform is compared, to making the operation that average waveform is adjusted to the amount of downward shift more loaded down with trivial details.Further, since production
Line (production line) is commonly used to manufacture non-defective unit, therefore obtains a number of normal waveform and be relatively easy to, but obtains
A number of unusual waveforms are comparatively more difficult, therefore the work to average waveform is adjusted to the amount of downward shift
Industry is extremely difficult.
Therefore, using method disclosed in patent document 1, the standard deviation of multiple normal waveforms is taken into account to determine to put down
Equal waveform m offset is more excellent in statistical significance, but as described above, according to method disclosed in patent document 1, then may be used
Excessive threshold value waveform can be generated.The problem of excessive threshold value waveform being generated using Figure 17 and Figure 18 explanations.
Figure 17 is the figure of the corresponding relation of multiple observation waveforms in the statistical disposition for schematically show patent document 1.Waveform
W1~w3 is the waveform of shape all same, but is slightly staggered on time-axis direction each other.Sight in waveform w1~w3 time t
The value of measured value direction of principal axis is expressed as f1 (t)~f3 (t).For example, the observation under moment tk is set to f1 (tk)~f3
(tk).The average and standard deviation of tk observation does not have this problem at the time of waveform w1~w3 flat parts, but right
Tj observation f1 (tj)~f3 (tj) average and standard deviation at the time of waveform w1~w3 ledge is observed,
Deviation on waveform w1~w3 time-axis direction obviously shows as the deviation of observation, and excessive threshold is generated so as to generate
The problem of being worth waveform.Because if for example, original situation, then in waveform w3 observation f3 (tj) and waveform w1
In observation f1 (ti) corresponding relation, rather than with observation f1 (tj) corresponding relation, the deviation should be by statistical appraisal.
In addition, if this existing statistical disposition is based on, respectively by the flat of f1 (tk)~fn (tk), f1 (tk)~fn (tk)
The value of equal waveform is expressed as the observation that m (tk) comes under the time tk to waveform w1~wn and carries out vague generalization, then average waveform energy
Obtain for
[mathematical expression 1]
Equally, f1 (tk)~fn (tk) standard deviation v (tk) can also obtain for
[mathematical expression 2]
Figure 18 is multiple observation waveform, average waveform, standard deviation and threshold values in the statistical disposition for represent patent document 1
The figure of waveform.The multiple waveform w4~w6 observed shown in Figure 18 (a) are to be of similar shape feature on bumps
Normal waveform, but be respectively provided with observation direction of principal axis and on time-axis direction and slightly offset.Figure 18 (b) is shown based on formula
(1) the average waveform m1 obtained after being averaged in each time to waveform w4~w6, but in average waveform m1, it is known that in waveform
Almost lose the convex portion as spike (overshoot) part after rising edge possessed by w4~w6.Because if root
According to formula (1), then the maximum of the respective convex portions of waveform w4~w6 is not averaged by average waveform m1, but such as the time 4
The average of waveform w4~w6 observation, waveform w4~w6 of time 5 observation it is average such, be averaged in each time
Processing.In addition, Figure 18 (c), which is shown, to amplify the value after specific times in the standard deviation that each time calculates based on formula (2), but
Standard deviation is amplified into the value after specific times as the larger value protruded in waveform w4~w6 rising edge or trailing edge part.
If generating threshold value waveform using the value amplified standard deviation after specific times and by the way of patent document 1, such as Figure 18
(d) shown in, it is known that generate leading edge portion or trailing edge part bound be known as excessive threshold value upper limit waveform h1 and
The threshold value waveform of lower limit waveform l1 defineds, from without representing correct upper and lower bound.
Next, the explanation carried out along Figure 16~Figure 18 utilized above, enters to the threshold value waveform generating of the present invention
Row explanation.Fig. 1 is the figure of the hardware configuration for the embodiment 1 for representing the threshold value waveform generating of the present invention.Threshold value waveform is given birth to
Include microprocessor (microprocessor) 2, system bus (system bus) 3, memory (storage into device 1
Memory) 4, input unit 5, storage unit 6 and display part 7.Memory 4 and storage unit 6 can generate the journey of threshold value waveform to performing
Sequence is stored.The program that microprocessor 2 generates threshold value waveform according to performing is handled, according to being stored in storage unit 6
Normal waveform generation average waveform and threshold value waveform.Display part 7 is used to confirm generated average waveform and threshold value waveform.Threshold
Value waveform generating 1 can be personal computer (personal computer), will be observed by check device or measurement apparatus
The normal waveform arrived preserves on a personal computer, the average wave generated on a personal computer in a manner of offline (offline)
Shape and threshold value waveform transfer use to check device.Threshold value waveform generating 1 can also be check device.In threshold value ripple
In the case that shape generating means 1 are check device, check device possesses the systematic function of average waveform and threshold value waveform, is based on
The normal waveform observed before is checked to generate average waveform and threshold value waveform and preserve, and judges the sight of check object accordingly
Whether good survey waveform.
Fig. 2 is the figure for representing to perform the structure of the program of the generation of threshold value waveform.Threshold value waveform generates program
(threshold wave shape creation program) 11 is stored in memory 4 or storage unit 6, including at noise
Reason portion (noise processing part) 12, feature extraction unit 13, relevant evaluation portion 14, statistical disposition portion 15 and threshold value
Waveform generating unit 16.
Threshold value waveform observes just in advance before generating program 11 first as needed in noise processed portion 12 to inspection
Ordinary wave shape carries out noise (noise) and removed.Because noise can be contained in the normal waveform observed to non-defective unit, therefore will
Extracted as characteristic point the part for preventing from causing script to be not wave character point because of the noise contained in normal waveform.Connect
, in feature extraction unit 13, the characteristic point of the normal waveform to observing is extracted.Then it is right in relevant evaluation portion 14
The characteristic point extracted by feature extraction unit 13 is evaluated, and using it is in characteristic point, have between multiple normal waveforms
The characteristic point of correlation is as true characteristic point.In statistical disposition portion 15, the observation of the true characteristic point of each normal waveform is calculated
Be averaged, also calculating the time is averaged.Here, the average computation of observation is one kind of the statistical disposition of observation direction of principal axis,
The average computation of time is one kind of the statistical disposition of time-axis direction.In addition, statistical disposition portion 15 is to the area between characteristic point
Between by drawing up to be consistent corresponding, come being averaged for calculating observation value and moment, the average wave that will be generated according to above-mentioned result of calculation
A normal waveform based on when shape is as generation threshold value waveform.In addition, statistical disposition portion 15 for calculating observation value and
The corresponding points calculating observation value of time mean time and the standard deviation of time.The standard deviation calculation of observation is observation axle
One kind of the statistical disposition in direction, the standard deviation calculation of time are one kind of the statistical disposition of time-axis direction.Finally, in threshold
Generation with the addition of observation and the standard of time to the average waveform obtained by statistical disposition portion 15 in value waveform generating unit 16
The threshold value waveform of deviation.The threshold value waveform that with the addition of the standard deviation of observation and time is upper limit waveform and lower notch
Shape.The multiple normal waveforms observed in advance the and characteristic point extracted, average waveform or threshold value waveform are carried out as needed
Compare, judge whether the generation of average waveform or threshold value waveform is reasonable, to involved by generation average waveform or threshold value waveform
Parameter (parameter) is adjusted, so as to generate average waveform or threshold value waveform again.
Noise processed portion 12 carries out noise remove (noise using low pass filter (lowpass filter)
rejection).The concrete example of low pass filter can enumerate moving average filter (moving average filter) or
Scale Space Filtering device (Scale Space Filtering).If, can using moving average filter as low pass filter
The degree of removable noise is adjusted according to time constant, if using Scale Space Filtering device as low pass filter, can
The degree of removable noise is adjusted according to yardstick.In addition, yardstick refers to the basic function i.e. Gaussian function of convolution integral
The range of (Gauss function).That is, can be joined in noise processed portion 12 by the removal to time constant or yardstick etc
Number 21 could be adjusted to carry out noise remove.
Although thus eliminating noise, the effect of smoothing may also cause the characteristic point for observing waveform to be lost, therefore,
Noise remove is not carried out preferably by excessive removal parameter (removal parameter) 21.Such as in moving average filtering
In the case of device, too big time constant should not be set, and bottom line should be limited in.However, if so, noise without
Method removes completely, so as to entirely prevent it being not that the part of wave character point is extracted as characteristic point originally.In order to answer
To this point, relevant evaluation portion 14 is provided with rear class.In addition, in the case of there is provided relevant evaluation portion 14, also should
The noise remove of bottom line is carried out in initial noise processed portion 12, to prevent that characteristic point from excessively being extracted, is caused originally
And the part of non-real characteristic point is be evaluated as being used as true characteristic point with correlation.
Feature extraction unit 13 extracts multiple characteristic points on waveform configuration respectively from the multiple normal waveforms observed.Ripple
Characteristic point in shape structure is, for example, the point that waveform configuration has large change, and is that the value for representing waveform intensity of variation exceedes in advance
The point of the threshold value first set.3~Fig. 7 of reference picture illustrates to this point.
Fig. 3 is to represent that waveform configuration has the figure of the point of large change.Fig. 3 shows an observed waveform,
Using waveform curvature as in the case of the value for representing waveform intensity of variation, on sampled point P1~P6, curvature has exceeded in advance
The threshold value of setting, thus sampled point P1~P6 turns into point of the waveform configuration with large change, i.e., the characteristic point on waveform configuration.
Further, since waveform is to link sampled point with line segment and form, therefore can also be to linking angle formed by the line segment of sampled point
Degree is evaluated to replace curvature, and point of the angle for the line segment for linking sampled point more than threshold value set in advance is considered as into waveform
Characteristic point in structure.
When extracting characteristic point by feature extraction unit 13, except setting the character references parameter (feature as threshold value
Standard parameter) 23, it is also necessary to set aspect ratio parameter (aspect ratio parameter) 22.Because
The transverse axis of waveform is the time, and the longitudinal axis is observation, and both units are different, therefore, reached according to observation relative to the time
Which kind of degree, the curvature of waveform and the value of angle can also produce change.
It is worth noting that, there is waveform configuration the judgement of the point of large change to need to carry out overall assessment to waveform.This
It is because if nearby judging the sampled point as object merely with threshold value, should be used as wave character point originally and carried
It may not be extracted as characteristic point the part taken.
Fig. 4 is the figure of the most simple types for the angle change for representing waveform.Fig. 4 is shown to be made up of sampled point P7~P14
Waveform.Carried using angle change as the value for representing waveform intensity of variation and to angle change progress threshold determination
In the case of taking characteristic point, the angle change of waveform can be defined as to such as angle change Δ θ1、Δθ2.Angle change Δ θ1
It is sampled point P10 angle change, the line segment equivalent to link sampled point P10, P11 is relative to the line for linking sampled point P9, P10
The angle of inclination of section.Angle change Δ θ2It is sampled point P11 angle change, equivalent to the line segment for linking sampled point P11, P12
Relative to the angle of inclination for the line segment for linking sampled point P10, P11.If using as the base of the benchmark for the characteristic point for judging concave shape
Quasi- angle, θ1It is set as+50 degree, using as the references angle θ of the benchmark for the characteristic point for judging convex form2It is set as -50 degree to make
Basic parameter 23 is characterized, then by angle change Δ θ1Sampled point P10 into+60 degree is partially formed as concave shape as waveform
Characteristic point and extract, by angle change Δ θ2It is partially formed as waveform into the point that -60 spend and is extracted for the characteristic point of convex form.
Even however, the feature identical waveform of concaveconvex shape, according to the difference of sample mode, concaveconvex shape is possible without being made
It is characterized a little and extracts.
Fig. 5 is the figure for representing the first kind needed for one group of angle change evaluation.Shown in waveform and Fig. 4 shown in Fig. 5
Waveform is equally made up of sampled point P7~P14, and the feature of concaveconvex shape is also identical, but the angle at sampled point P9, P10 point
Changes delta θ3、Δθ1Not less than the references angle θ shown in Fig. 41, thus sampled point P9, P10 not feature as concave shape
Point is extracted.In addition, angle change Δ θ3It is sampled point P9 angle change, equivalent to the line segment phase for linking sampled point P9, P10
For the angle of inclination for the line segment for linking sampled point P8, P9.
Fig. 6 is the figure for representing the Second Type needed for one group of angle change evaluation.Shown in waveform and Fig. 4 shown in Fig. 6
Waveform is equally made up of sampled point P7~P14, and the feature of concaveconvex shape is also identical, but the angle at sampled point P11, P12 point
Spend changes delta θ2、Δθ4Not less than the references angle θ shown in Fig. 42, thus sampled point P11, P12 not spy as convex form
Sign point is extracted.In addition, angle change Δ θ4It is sampled point P12 angle change, equivalent to the line for linking sampled point P12, P13
Section relative to link sampled point P11, P12 line segment angle of inclination.
In Fig. 5 and Fig. 6, for the situation for preventing characteristic point to be not extracted by, continuous angle change in the same direction is tackled
Point group be uniformly processed, using the summation of the angle change of point group after reunification as one group of angle change, and to this progress
Threshold determination.Such as in the case of Fig. 5 waveform, handled using sampled point P9, P10 as one group, if judging to be used as one group
Sampled point P9, P10 angle change Δ θ1' (=angle change Δ θ3+ angle change Δ θ1) exceed references angle θ1, then will
Point group is that the point of sampled point P9, P10 right in the middle is drawn up and handled for the characteristic point of concave shape.Also it is same in the case of Fig. 6
Sample, handled using sampled point P11, P12 as one group, if judging the angle change Δ θ of sampled point P11, P12 as one group2′
(=angle change Δ θ2+ angle change Δ θ4) exceed references angle θ2, then it is the point of sampled point P11, P12 right in the middle by point group
Draw up and handled for the characteristic point of convex form.
Fig. 7 is the figure for the relation for schematically showing the character pair point between different wave.Fig. 7 show two waveform W1,
W2, the observation of waveform W1, W2 in time T are represented with F1 (T), F2 (T) respectively.It is right with observation F1 (T2) on waveform W1
Sampled point P21 corresponding with observation F2 (T3) is the starting point of rising edge on the sampled point P15 and waveform W2 answered, in difference
Moment is extracted as mutually corresponding characteristic point.Sampled point P16 and waveform corresponding with observation F1 (T3) on waveform W1
The upper sampled point P22s corresponding with observation F2 (T4) of W2 is the end point of rising edge, at different moments as mutually corresponding special
Sign point is extracted.Then, on waveform W1 on sampled point P19 and waveform W2 corresponding with observation F1 (T7) with observation F2
(T6) sampled point P24 corresponding to is the starting point of trailing edge, is being extracted at different moments as mutually corresponding characteristic point.Waveform
Sampled point P25 corresponding with observation F2 (T7) is on the upper sampled point P20 and waveform W2 corresponding with observation F1 (T8) of W1
The end point of trailing edge, it is being extracted at different moments as mutually corresponding characteristic point.Moreover, it is assumed that the area between each characteristic point
Between unanimously correspond to, so as to by sampled point P18 corresponding with observation F1 (T5) waveform W1 on draw up for on waveform W2 and observe
Point corresponding sampled point P23 corresponding to value F2 (T5).In addition, by sampled point corresponding with observation F1 (T4) on waveform W1
P17 draws up as the point corresponding with the sampled point P22 on waveform W2,23 midpoint.
Therefore, by extracting the characteristic point of normal waveform in advance by feature extraction unit 13, so as to such as illustrated in fig. 7
Point corresponding to being determined like that between multiple normal waveforms.Therefore, can be in observation direction of principal axis in threshold value waveform generates program
With the statistical appraisal that average or standard deviation etc is carried out on the two directions of time-axis direction.Such as in the case of fig. 7, ripple
The starting point of the rising edge of shape W1, W2 be the observation at sampled point P15, P21 average value M (Ti) by
[mathematical expression 3]
[mathematical expression 4]
To try to achieve, the standard deviation of normal waveform is set to standard deviation Vf (Ti) and the time shaft side of observation direction of principal axis
To standard deviation Vt (Ti), respectively by
[mathematical expression 5]
[mathematical expression 6]
To try to achieve.
Relevant evaluation portion 14 is using the characteristic point conduct in the characteristic point extracted from multiple normal waveforms with correlation
True characteristic point, and by true characteristic point packet (grouping) into multiple so that same group (group) is by with identical correlation
Multiple characteristic points form.This is to prevent from causing because of influence of noise not being the part of wave character point originally by as spy
Levy point and extract.Because multiple normal waveforms have a mutually the same wave character, therefore from the substantially phase of each normal waveform
Same position extracts characteristic point.Also, only from a part of waveform extracting to characteristic point can be construed to because of influence of noise and
The a reference value of feature judgement is exceeded.Reference picture 8 illustrates to this point.
Fig. 8 is the figure for representing the characteristic point and its correlation extracted by feature extraction unit.Fig. 8 (a) show waveform W3 and
Waveform W3 characteristic point P26~P30, Fig. 8 (b) shows that waveform W4 and waveform W4 characteristic point P31~P36, Fig. 8 (c) shows ripple
Shape W5 and waveform W5 characteristic point P37~P42, Fig. 8 (d), which show only to draw, (plot) these characteristic points P26~P42 shape
State.Understand near Fig. 8 (d) time 3~5, recessed characteristic point P26, P31, P37 for being extracted from waveform W3~W5 connect each other
Closely.Thus, the characteristic point of the same race close within the specific limits extracted from any one in waveform W3~W5 all thinks have
Correlation, therefore used by relevant evaluation portion 14 as true characteristic point.In addition, characteristic point of the same race here refers to that identical is recessed
Characteristic point or the convex characteristic point of identical.In addition, relevant evaluation portion 14 is to having identical correlation between above-mentioned multiple waveform W3~W5
Characteristic point P26, P31, P37 of property be grouped, and is used as the point group to correspond to each other.Setting or adjustment as judge with
The threshold value of the close and related judgment standard of the scope of which kind of degree is used as degree of correlation parameter (correlation
parameter)24.For example, the concrete example of degree of correlation parameter 24 can enumerate time-axis direction close to scope and observation axle
Direction close to scope.Thus, in relevant evaluation portion 14, in the convex feature that the time 4~6 nearby extracts from waveform W3~W5
Point P27, P32, P38 are characteristic point of the same race closer to each other, so as to be judged as with correlation, therefore are adopted as true feature
Point, and it is classified as one group as the point group with identical correlation.In addition, in relevant evaluation portion 14, near the time 5~8
Recessed characteristic point P28, P33, the P39 extracted from waveform W3~W5 is characteristic point of the same race closer to each other, so as to be judged as having
Correlation, therefore true characteristic point is adopted as, and it is classified as one group as the point group with identical correlation.In addition, in phase
Close in evaluation section 14, be nearby to connect each other from waveform W3~W5 convex characteristic point P29, P35, P41 extracted in the time 23~25
Near characteristic point of the same race, so as to be judged as with correlation, therefore is adopted as true characteristic point, and conduct has identical correlation
The point group of property and be classified as one group.In addition, in relevant evaluation portion 14, nearby extracted in the time 25~27 from waveform W3~W5
Recessed characteristic point P30, P36, P42 are characteristic point of the same race closer to each other, so as to be judged as with correlation, therefore are adopted as
True characteristic point, and it is classified as one group as the point group with identical correlation.In addition, recessed characteristic point P26 near the time 3~5,
P31, P37 and recessed characteristic point P28, P33, P39 near the time 5~8 are recessed characteristic points of the same race, although the moment approaches, by
Kept off in observation, therefore relevant evaluation portion 14 is not judged as with correlation.That is, not by characteristic point P26, P31, P37,
P28, P33, P39 are grouped into the point group with identical correlation.In addition, the convex spy only extracted near the time 15 from waveform W4
Proximity relation is not present in sign point P34 and convex characteristic point P27, P29, P38, P41 for being extracted from other waveform W3, W5, therefore
True characteristic point is not adopted as.Equally, the recessed characteristic point P40 only extracted near the time 16 from waveform W5 also with from other
Proximity relation is not present in recessed characteristic point P26, P28, P30, P31, P33, P36 that waveform W3, W4 are extracted, therefore is not used
As true characteristic point.
Statistical disposition portion 15 calculates according to group to be employed as true characteristic point by relevant evaluation portion 14 and is grouped to obtain
Each normal waveform the observation of characteristic point and being averaged for time, so as to generate average waveform.In addition, by feature
Drawn up in section between point to be consistent corresponding, come point corresponding to determining, according to the corresponding points calculating observation of each normal waveform
Value and time are averaged.In addition, on true characteristic point, although the corresponding points as sampled point on each normal waveform be present,
But following situation is also had sometimes：That is, in the section between characteristic point, the point corresponding with sampled point on a normal waveform
Sampled point is not intended as on other normal waveforms and is existed, and is existed as the point on line segment, accordingly it is also possible to calculate this
Point on line segment, can also by the use of existing sampled point near the point on line segment come instead of and approximate be used as corresponding points.
If the corresponding points between normal waveform can be determined as described above, average wave can be obtained based on formula (3) and formula (4)
Shape, turn into basic normal waveform during as generation threshold value waveform.Statistical disposition portion 15 is also based on formula (5) and formula (6)
The standard deviation of calculating observation value direction of principal axis and time-axis direction respectively.Normal waveform by n sampling point group form, if by when
Between be set to sample number, then the time of each sampled point is integer value, but because the time of each point of resulting average waveform is
The result of average calculating operation, therefore be generally not integer value.As shown in figure 16, in order to carry out waveform comparison, average waveform and threshold
Value waveform is also required to be made up of n sampling point group, and the time of each sampled point is also required to match.Therefore, average calculating operation can be made
Result be integer and as average waveform, the line that the point group represented by the coordinate value of fractional value can also be formed
Section enters row interpolation, calculates the point group represented by the coordinate value of integer value, and it is averaged.As the knot for making average calculating operation
Fruit is an example of the method for integer, such as is had using rounding up come the method rounded.
Fig. 9 is the figure for the standard deviation for representing the average waveform of embodiment 1, the standard deviation of observation and time.
Fig. 9 (a) shows each normal waveform for being adopted as true characteristic point according to Fig. 8 (d) in statistical disposition portion 15
True characteristic point, calculate putting down for observation based on formula (3) and formula (4) and time for each point group with identical correlation
The average waveform M1 generated.Average waveform M1 characteristic point P43 is characteristic point P26, P31, P37 for representing Fig. 8 (d) sight
The point of the average value of measured value and the average value of time and be with by characteristic point P26, P31, P37 structure with identical correlation
Into the corresponding point of point group.Equally, average waveform M1 characteristic point P44 is characteristic point P27, P32, the P38 for representing Fig. 8 (d)
The average value of observation and the point of the average value of time and be with by characteristic point P27, P32 with identical correlation,
The corresponding point of point group that P38 is formed.Characteristic point P45 is the flat of characteristic point P28, P33, P39 for representing Fig. 8 (d) observation
The point of the average value of average and time and be the point group formed with by characteristic point P28, P33, P39 with identical correlation
Corresponding point.Characteristic point P46 is average value and the time of characteristic point P29, P35, P41 for representing Fig. 8 (d) observation
The point of average value and be the corresponding point of the point group formed with by characteristic point P29, P35, P41 with identical correlation.Feature
Point P47 is the average value of characteristic point P30, P36, P42 for representing Fig. 8 (d) observation and the point of the average value of time and is
The corresponding point of the point group that is formed with by characteristic point P30, P36, P42 with identical correlation.Fig. 9 (b), which is shown, to be counted
The true characteristic point of each normal waveform for being adopted as true characteristic point in processing unit 15 according to Fig. 8 (d), for each
The standard deviation of point group with identical correlation, the observation calculated based on formula (5).Fig. 9 (c) is shown in statistical disposition
The true characteristic point of each normal waveform for being adopted as true characteristic point in portion 15 according to Fig. 8 (d), has for each
The point group of identical correlation, the standard deviation of the time calculated based on formula (6).Statistical disposition shown in Fig. 9 (a)~(c)
Value shows the time deviation of the waveform in the existing method shown in Figure 18 in standard deviation eventually as observation deviation
Situation is evaluated in the form of the deviation of observation direction of principal axis and time-axis direction.
Characteristic point of the threshold value waveform generating unit 16 according to permissibility set in advance to the average waveform M1 shown in Fig. 9 (a)
P43~P47 add respectively according to characteristic point P43~P47 respectively corresponding to there is the every of identical correlation shown in Fig. 8 (d)
The standard deviation for the observation that individual point group calculates and the standard deviation of time, so as to generate threshold value waveform.In addition, generated
Threshold value waveform be upper limit waveform and lower limit waveform.Standard deviation and the time of observation are added according to permissibility set in advance
Standard deviation refer to the amount for adding the certain multiple of observation or the standard deviation of time, the certain multiple added is by allowing
Parameter (tolerance parameter) 25 is spent to set or adjust.Reference picture 10 and Figure 11 illustrate to this point.
Figure 10 is the standard deviation and the mark of time for representing the average waveform addition observation to being calculated in embodiment 1
Quasi- deviation and the figure of threshold value waveform generated.Figure 11 is the evaluation method for schematically illustrating average waveform in threshold value waveform generating unit
Figure.Figure 10 (a) is shown at average waveform M1 characteristic point P43~P47 by the standard deviation of observation and the mark of time
The ellipse that quasi- deviation is formed.Here, when showing that 3 times of the standard deviation by observation and time are set as permissibility parameter 25
Ellipse.If oval viewpoint of the threshold value waveform generating unit 16 according to Figure 10 (a) is evaluated average waveform M1, can
The outer peripheral waveform for forming these oval groups is defined as upper limit waveform H1 and lower limit waveform as shown in Figure 10 (b)
L1。
Here, one of constructive method as outer peripheral upper limit waveform H1 and lower limit the waveform L1 for forming oval group
Example, such as have：The space surrounded by oval group is set to be determined as good space, average waveform is evaluated, and root
The upper limit waveform and lower limit waveform as threshold value waveform are generated according to average waveform.Specifically, Tu11Zhong, by shown in each time
The scopes that are surrounded of oval C1~C3 be set to that good space should be determined as.Oval C1 is with the average waveform equivalent to Figure 10 (a)
Length of the 3 times of values of the standard deviation of the observation corresponding to time T10 in M1 average waveform as short axle A1, with average wave
Length of the 3 times of values of the standard deviation of the time corresponding to time T10 in shape as major axis A 2.Oval C2 is with average waveform
Length of the 3 times of values of the standard deviation of the observation corresponding to time T11 as short axle A3, with the time T11 institutes in average waveform
Length of the 3 times of values of the standard deviation of corresponding time as major axis A 4.Oval C3 is with corresponding to the time T12 in average waveform
Length of the 3 times of values of the standard deviation of observation as major axis A 5, with the standard of the time corresponding to the time T12 in average waveform
Length of the 3 times of values of deviation as short axle A6.As upper limit waveform and the concrete example of the generation method of lower limit waveform, can enumerate
As shown in Figure 11, to short axle A1, A3, A6 or length corresponding to the point plus-minus in the average waveform corresponding to each time T10~T12
Axle A2, A4, A5 length so that average waveform is offset to observation direction of principal axis and time-axis direction.For example, in the ellipse of Figure 11
In circle C1, if short axle A1 length is added to the observation corresponding to the time T10 of average waveform, and to the time of average waveform
Time corresponding to T10 subtracts the length of major axis A 2, then as the point on upper limit waveform.Equally, in oval C1, if to average
Observation corresponding to the time T10 of waveform subtracts short axle A1 length, and the time corresponding to by the time T10 of average waveform
Plus the length of major axis A 2, then as the point on lower limit waveform.That is, in method disclosed in patent document 2, it will be observed that it is more
The space that individual normal waveform is surrounded is set to that good space should be determined as to generate threshold value waveform, and in embodiment 1, will be by
The space that the oval group that the standard deviation of standard deviation and moment for the observation of average waveform each point is formed is surrounded
It is set to that good space should be determined as to generate threshold value waveform.
During so far, the parameter to be adjusted, which has, removes parameter 21, aspect ratio parameter 22, character references parameter
23rd, degree of correlation parameter 24, permissibility parameter 25 these parameters, by being set to them, resulting average waveform M1
And threshold value waveform would also vary from.In addition, threshold value waveform mentioned here refers to upper limit waveform H1 and lower limit waveform L1.Cause
This, can be using the characteristic point P43~P47 extracted, average waveform M1, upper limit waveform H1 and lower limit waveform as threshold value waveform
Whether L1 judges upper limit waveform H1 and lower limit waveform L1 compared with the multiple normal waveforms i.e. waveform W3~W5 observed
Rationally, and above-mentioned parameter can be adjusted as needed, so as to generate average waveform M1 and threshold value waveform again.It is such as right
Parameter is adjusted, so as to which as shown in Figure 10 (c), the multiple normal waveforms for making to observe i.e. waveform W3~W5 is limited in resulting
Threshold value waveform be in upper limit waveform H1 and lower limit waveform L1.In addition, when setting or adjusting above-mentioned parameter, can be to whole
Wave setting simultaneously applies single constant value, can also set constant value different under each time and be finely adjusted.
Thus, can be on observation direction of principal axis and the two directions of time-axis direction to resulting according to embodiment 1
Multiple normal waveforms are that waveform W3~W5 deviation is evaluated, thus can obtain a really average average waveform M1 without
The feature of waveform can be lost because of the smooth effect that equalization is brought, moreover, the deviation of time will not also be presented as the inclined of observation
Difference, offset can be determined using the standard deviation of observation direction of principal axis and the standard deviation in moment time direction, and can generated
It with the addition of threshold value waveform, the i.e. upper limit waveform H1 and lower limit waveform L1 of real deviation.It is also, real by using with the addition of
The threshold value waveform of deviation, can realize just right waveform comparison.
In addition, the quality based on threshold value waveform judges to can be used not only for product examination used by embodiment 1, can also use
In supervisor control (monitoring control system) abnormality detection.Exception as supervisor control is examined
The example surveyed, can enumerate unexpected alarm or the police of daily power consumption in social infrastructure (infrastructure)
Report.In existing supervisor control, as the judgment standard for abnormality detection, only prepare to utilize higher limit, lower limit
The unified judgement of value, upper higher limit, lower lower limit, and if using this method, such as can be according to daily power consumption mode
Normal power consumption mode generation threshold value waveform in (electricity consumption pattern), so as to more meticulously enter
Whether row saves the judgement of (energy saving), can prevent from wasting caused by alarm in advance.According to normal power consumption mould
In the case of formula generation threshold value waveform, carved at the beginning of power consumption and finish time also can increasingly offset, therefore with utilizing existing side
The judgement that the threshold value waveform of method generation is carried out is compared, and the threshold value waveform generated using embodiment 1 can be by than ever more
It is rational to judge to obtain alarm.
Embodiment 2.
In embodiment 1, in order that the evaluation of estimate that statistical disposition as average and standard deviation is related to is reasonable enough,
Need to observe a number of normal waveform.Therefore, in embodiment 2, even if illustrating not pair so that what statistical disposition was related to
Evaluation of estimate is reasonable enough and the normal waveform of requirement is observed, also can be by system as average and standard deviation
The involved evaluation of estimate of meter processing is calculated to generate the method for average waveform and threshold value waveform.In embodiment 2, to see
In case of the normal waveform measured is one, threshold value waveform generation program 11 is calculated as average and standard deviation
Evaluation of estimate involved by statistical disposition illustrates to generate the situation of average waveform and threshold value waveform.
In embodiment 1, in order to prevent because the noise being superimposed upon in waveform causes originally the not part of wave character point
Characteristic point is extracted as, noise processed is carried out in noise processed portion 12, on the other hand, in order to prevent wave character point because making an uproar
Sound removes the smooth effect brought and lost, and is limited in and carries out noise remove using the removal parameter 21 of bottom line, in correlation
Only it is associated in evaluation section 14 using the characteristic point with correlation as true characteristic point.However, in an only normal waveform
In the case of, correlation can not be evaluated in relevant evaluation portion 14 as Embodiment 1.Therefore, in embodiment 2, with
A certain degree of larger removal parameter 21 carries out noise remove, and gradually changes the character references parameter 23 when extracting characteristic point
To tackle.If gradually increasing smoothness, the noise of high fdrequency component is removed first, and the feature structure of waveform originally is being made an uproar
Sound is gradually lost after being removed, therefore larger removal parameter 21 is used in the degree that can remove noise, for not losing also
Wave character, the threshold value for reducing character references parameter 23 extracted.
Figure 12 is the figure for the feature point extraction based on smoothness adjustment for representing embodiment 2.One observed is just
Noise is included in ordinary wave shape W6.Scale Space Filtering (filtering) is used in embodiment 2 as the method for removing noise.
With Scale Space Filtering mesoscale s increase, waveform W6 general configuration is left behind, therefore using yardstick s as noise remove
Parameter (noise rejection parameter) is adjusted.To a waveform W6, by by yardstick s be set to yardstick s=1,
The filtered each waveform W61 of Scaling interval, waveform W62, waveform W63 when yardstick s=2, yardstick s=3 are progressively removed and made an uproar
Sound, if the waveform W63 as corresponding to 13 pairs of feature extraction unit with yardstick s=3 extracts characteristic point, convex characteristic point P49, P51 with
And recessed characteristic point P48, P50, P52 are extracted, using characteristic point P48~P52 as true characteristic point in relevant evaluation portion 14.Separately
Outside, on Scale Space Filtering, such as non-patent literature (Scale Space Filtering (the electronic information communication association paper of periodic waveform
Collect D Vol.J73-D2No.4pp.544-552, issue date 1990/04/25)) it is disclosed as, demonstrate with yardstick s's
Increase, waveform W6 structure lose dullness.
In the case of an only normal waveform, correlation can not be not only evaluated, can not also evaluate average and standard deviation
Difference.Therefore, in relevant evaluation portion 14, the feature extracted to the waveform W63 that noise remove has been carried out with yardstick s=3 is used
Characteristic point P53~P57 on waveform W6 before point P48~P52 noise remove determines as the true characteristic point on waveform W6
The flat section in section between true characteristic point P53~P57.That is, the distinguished point based P48~P52 of relevant evaluation portion 14 comes true
Standing wave shape W6 flat section.The degree of correlation parameter 24 of embodiment 2 is to flat as the inclination of which kind of degree is judged as
The threshold value of benchmark is set or adjusted.For example, if two observations in characteristic point P53~P57 in threshold range, are incited somebody to action
Interval judgement between two characteristic points is flat, or using least square method etc. to two positions in characteristic point P53~P57
Point group between putting carries out straight line approximation, if the slope of the straight line after approximation in threshold range, is judged as flat.Also,
In statistical disposition portion 15, handled using a normal waveform W6 as average waveform.
Figure 13 is the standard for schematically showing the observation direction of principal axis that the point group based on the observation in flat section is carried out
The figure of the evaluation of deviation.Statistical disposition portion 15 belongs to flat section T by what is determined by relevant evaluation portion 14jkPoint group as pair
As the statistical disposition only on observation direction of principal axis, performing following calculating, i.e.,：Calculating belongs to by institute of relevant evaluation portion 14 really
Fixed flat section TjkPoint group corresponding to the deviation of observation be standard deviation Vf.That is, when average waveform, i.e. normal ripple
T observation is represented by F (T) at the time of shape W6, and normal waveform W6 flat section TjkObservation be from F (Tj) to F
(TK) when, the average value M of observationjkCan be by
To try to achieve, standard deviation VfjkCan be by
To try to achieve.Standard deviation VfjkAs flat section TjkThe standard deviation of observation direction of principal axis of each point locate
Reason.Part flat waveform W6 should be in the ideal case flat straight line, but in addition to the noise of superposition, observation pair
The fluctuation of the physical phenomenon of elephant in itself, which also results in, there is deviation in observation, therefore, even if not to the observation of multiple waveforms
Deviation carry out statistical appraisal, replace and statistical appraisal carried out to the deviation of the observation of the flat of same waveform,
Its purport is not departed from.Moreover, the standard deviation Vf that formula (7) and formula (8) calculate will be based onjkDraw up as flat section TjkRespectively adopt
The standard deviation of time-axis direction at sampling point, also draw up and be characterized between point P53~P57, except flat section TjkIn addition its
The standard deviation of observation direction of principal axis and the standard deviation of time-axis direction of each sample point in its section is handled.Since
Normal waveform be waveform W6 this, then for flat section TjkObservation direction beyond deviation, do not find and do not depart from
The suitable computational methods of its purport, therefore it is consistent to draw up as deviation in the range of whole waveform W6.
Thus, the standard deviation of the observation of the waveform W6 as average waveform and each point on waveform W6 can be tried to achieve
And the standard deviation of time, therefore, threshold value waveform can be equally generated with embodiment 1 in threshold value waveform generating unit 16.
In addition, it is two or more but negligible amounts, causes based on system as average and standard deviation in resulting normal waveform
Count the evaluation of estimate of processing and be insufficient to it is rational in the case of, by a small number of according to these but still be that multiple normal waveforms utilize
The statistic that embodiment 1 calculates is i.e. average and standard deviation and each normal waveform extrapolated using embodiment 2
Statistic it is i.e. average and standard deviation carries out average or weighted average, so as to obtain a statistic, according to this system
Metering can also generate threshold value waveform.
Thus,, also can be by based on pushing away even if resulting normal waveform is one or a small number of according to embodiment 2
Evaluating to replace the statistic evaluated in embodiment 1 for calculation, threshold value is generated so as to utilization and the identical method of embodiment 1
Waveform.
Embodiment 3.
In embodiment 1 and embodiment 2, the situation that threshold value waveform is generated according to normal waveform is described,
But in previously known abnormal producing cause and in the case of unusual waveforms caused by reason of the same race can be observed, sometimes need not
To whole observation wave setting threshold value waveform, and only need to be to producing given threshold waveform near exception.Embodiment 3
In, this threshold value waveform is set for illustrating.
Figure 14 is the figure of the structure for the embodiment 3 for representing the threshold value waveform generating of the present invention.In embodiment 3,
New in threshold value waveform generation program 11 to be provided with abnormal generation section extraction unit 33, thus, threshold value waveform generating unit 16 possesses only
Function to producing given threshold waveform near exception in observation waveform.Figure 15 is schematically shown in embodiment 3 to different
The figure of the generating process of section generation threshold value waveform is often produced, Figure 15 (a) shows the threshold value waveform generation journey based on embodiment 3
Sequence 11 and the average waveform M2 calculated, Figure 15 (b) show it is abnormal produce section S2, Figure 15 (c) shows only to produce section to abnormal
Threshold value waveform, the i.e. upper limit waveform H2 and lower limit waveform L2 of S2 generations.
In embodiment 3, threshold value waveform generates program 11 using noise processed portion 12 respectively to being observed in advance before inspection
Normal waveform 31 or unusual waveforms 32 carry out necessary noise remove, characteristic point is extracted by feature extraction unit 13, commented by correlation
Valency portion 14 and statistical disposition portion 15 carry out relevant evaluation and statistical disposition, calculate average waveform M2 and standard deviation.That is, exist
In the case of multiple normal waveforms 31 being observed, by with the identical of embodiment 1 in a manner of evaluate correlation, and extract true feature
Point, in the case where being only capable of using normal waveform, by with the identical of embodiment 2 in a manner of extract characteristic point.In addition, in energy
In the case of observing multiple unusual waveforms 32, by and the identical of embodiment 1 in a manner of by between above-mentioned multiple unusual waveforms 32 that
There is the characteristic point of correlation to be extracted as true characteristic point for this, in the case where being only capable of using a unusual waveforms 32, with
Characteristic point is extracted with the identical mode of embodiment 2.
Then, in section extraction unit 33 is produced extremely, the characteristic point of normal waveform 31 and the spy of unusual waveforms 32 are evaluated
Correlation between sign point, exception is regarded as in the section for forming the characteristic point by no correlation as shown in Figure 15 (b)
Produce section S2.Which kind of now, adjusted using abnormal basic parameter (abnormal standard parameter) 34 with journey
The width of degree is abnormal generation section S2 to regard as.For example, as shown in Figure 15 (b), considered in the positive direction of time shaft and
In negative direction, characteristic point P58~P61 defineds pair with the unusual waveforms 32 of the characteristic point of normal waveform 31 without correlation
Section S 1, plus the standard deviation V of the time corresponding with characteristic point P58~P61 of unusual waveforms 32 3 times of value after
Width regard as abnormal producing section S2.In the case where being only capable of using an abnormal generation waveform 32, can be based on passing through
The standard deviation that embodiment 2 is drawn up, but can also be existed based on the characteristic point in the section of the normal waveform 31 equivalent to section S1
The standard deviation of time-axis direction.That is, in characteristic point and unusual waveforms 32 based on the normal waveform 31 each other with correlation
Characteristic point, each point drawn up as in the case of corresponding to consistent, based on the characteristic point of normal waveform 31 without correlation
The standard deviation of characteristic point P58~P61 times of corresponding points on normal waveform 31 of unusual waveforms 32.
Finally, as shown in Figure 15 (c), threshold value waveform generating unit 16 only to thus obtained abnormal generation section S2, utilizes
The average waveform M2 and standard deviation calculated by statistical disposition portion 15 come generate as threshold value waveform upper limit waveform H2 and under
Notch shape L2.
According to embodiment 3, without judging bound to the whole waveform observed, judge that required processing time obtains
To shorten.Requiring high efficiency, a Product processing was shortened, so that must to the time between next Product processing
It must terminate that there is practical significance in the case of determination processing in a short time.
Industrial practicality
As described above, the present invention threshold value waveform generating can obtain with the addition of observation waveform time-axis direction it is inclined
Difference will not appear as the threshold value waveform of the true deviation of the deviation of observation direction of principal axis, so as to the addition of true deviation using
Threshold value waveform this aspect is useful.
Label declaration
1 threshold value waveform generating
2 microprocessors
3 system bus
4 memories
5 input units
6 storage units
7 display parts
11 threshold value waveforms generate program
12 noise processed portions
13 feature extraction units
14 relevant evaluation portions
15 statistical disposition portions
16 threshold value waveform generating units
21 remove parameter
22 aspect ratio parameters
23 character references parameters
24 degree of correlation parameters
25 permissibility parameters
31 normal waveforms
32 unusual waveforms
33 abnormal generation section extraction units
34 abnormal basic parameters
P1~P25 sampled points
P26~P61 characteristic points
θ1、θ2References angle
Δθ1~Δ θ4Angle change
Δθ1′、Δθ2' mono- group of angle change
W1~W6, w1~w6 waveforms
M1, M2, m, m1 average waveform
H1, H2, h, h1 upper limit waveform
L1, L2, l, l1 lower limit waveform
C1~C3 is oval
A1, A3, A6 short axle
A2, A4, A5 major axis
S yardsticks
TjkFlat section
S1 sections
S2 produces section extremely.
Claims (15)
1. a kind of threshold value waveform generating, generate the threshold value waveform of the normality for judging observation waveform, it is characterised in that
Including：
Feature extraction unit, this feature extraction unit extract characteristic point from observation waveform；
Relevant evaluation portion, the relevant evaluation portion are evaluated the characteristic point extracted by the feature extraction unit；
Statistical disposition portion, the statistical disposition portion are observed the time-axis direction and observation axle of waveform based on the characteristic point
The statistical disposition in direction；And
Threshold value waveform generating unit, the threshold value waveform generating unit is based on the time-axis direction and the observation direction of principal axis
Statistical disposition generates the threshold value waveform,
The feature extraction unit extracts multiple characteristic points respectively from multiple observation waveforms,
The characteristic point is grouped into multiple by the relevant evaluation portion so that same group is by multiple spies with identical correlation
Sign point is formed,
As the statistical disposition, the statistical disposition portion calculates the observation of the characteristic point according to each group respectively
It is average and the time be averaged to generate average waveform, and calculate the standard deviation of the observation and the time,
Point in the threshold value waveform generating unit pair average waveform corresponding with the group adds the standard deviation
To generate the threshold value waveform.
2. threshold value waveform generating as claimed in claim 1, it is characterised in that the threshold value waveform generating unit is to described flat
The point on equal waveform adds and subtracts the value of the observation and the standard deviation of the time based on the corresponding group,
Thus upper limit waveform and lower limit waveform as the threshold value waveform are generated.
3. a kind of threshold value waveform generating, generate the threshold value waveform of the normality for judging observation waveform, it is characterised in that
Including：
Feature extraction unit, this feature extraction unit extract characteristic point from observation waveform；
Relevant evaluation portion, the relevant evaluation portion are evaluated the characteristic point extracted by the feature extraction unit；
Statistical disposition portion, the statistical disposition portion are observed the time-axis direction and observation axle of waveform based on the characteristic point
The statistical disposition in direction；And
Threshold value waveform generating unit, the threshold value waveform generating unit is based on the time-axis direction and the observation direction of principal axis
Statistical disposition generates the threshold value waveform,
The feature extraction unit extracts the characteristic point from an observation waveform,
The relevant evaluation portion determines the flat section of one observation waveform based on the characteristic point,
The statistical disposition portion calculates the standard deviation of the observation corresponding with belonging to multiple point groups in the flat section, will
The standard deviation of the observation calculated is drawn up as the standard deviation of the time in the flat section, and is drawn up as institute
The observation and the standard deviation of time in other sections in section, in addition to the flat section are stated,
The threshold value waveform generating unit adds the observation and the standard deviation of the time to one observation waveform
To generate the threshold value waveform.
4. a kind of threshold value waveform generating, generate the threshold value waveform of the normality for judging observation waveform, it is characterised in that
Including：
Feature extraction unit, this feature extraction unit extract characteristic point from observation waveform；
Relevant evaluation portion, the relevant evaluation portion are evaluated the characteristic point extracted by the feature extraction unit；
Statistical disposition portion, the statistical disposition portion are observed the time-axis direction and observation axle of waveform based on the characteristic point
The statistical disposition in direction；And
Threshold value waveform generating unit, the threshold value waveform generating unit is based on the time-axis direction and the observation direction of principal axis
Statistical disposition generates the threshold value waveform,
Section extraction unit is produced including abnormal, this produces section extraction unit by the spy of the normal waveform in the observation waveform extremely
The area that the characteristic point without correlation is formed between sign point and the characteristic point of the unusual waveforms in the observation waveform
Between regard as it is abnormal produce section,
The threshold value waveform generating unit only generates the threshold value waveform to the abnormal section that produces.
5. threshold value waveform generating as claimed in claim 1, it is characterised in that section extraction unit is produced including abnormal, should
The abnormal section extraction unit that produces is by the characteristic point of the normal waveform in the observation waveform and the extraordinary wave in the observation waveform
Abnormal generation section is regarded as in the section that the characteristic point without correlation is formed between the characteristic point of shape,
The threshold value waveform generating unit only generates the threshold value waveform to the abnormal section that produces.
6. threshold value waveform generating as claimed in claim 2, it is characterised in that section extraction unit is produced including abnormal, should
The abnormal section extraction unit that produces is by the characteristic point of the normal waveform in the observation waveform and the extraordinary wave in the observation waveform
Abnormal generation section is regarded as in the section that the characteristic point without correlation is formed between the characteristic point of shape,
The threshold value waveform generating unit only generates the threshold value waveform to the abnormal section that produces.
7. threshold value waveform generating as claimed in claim 3, it is characterised in that section extraction unit is produced including abnormal, should
The abnormal section extraction unit that produces is by the characteristic point of the normal waveform in the observation waveform and the extraordinary wave in the observation waveform
Abnormal generation section is regarded as in the section that the characteristic point without correlation is formed between the characteristic point of shape,
The threshold value waveform generating unit only generates the threshold value waveform to the abnormal section that produces.
8. threshold value waveform generating as claimed in claim 1, it is characterised in that the feature extraction unit will represent the sight
The value for surveying the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
9. threshold value waveform generating as claimed in claim 2, it is characterised in that the feature extraction unit will represent the sight
The value for surveying the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
10. threshold value waveform generating as claimed in claim 3, it is characterised in that described in the feature extraction unit will represent
The value for observing the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
11. threshold value waveform generating as claimed in claim 4, it is characterised in that described in the feature extraction unit will represent
The value for observing the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
12. threshold value waveform generating as claimed in claim 5, it is characterised in that described in the feature extraction unit will represent
The value for observing the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
13. threshold value waveform generating as claimed in claim 6, it is characterised in that described in the feature extraction unit will represent
The value for observing the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
14. threshold value waveform generating as claimed in claim 7, it is characterised in that described in the feature extraction unit will represent
The value for observing the intensity of variation of waveform is extracted more than the point of threshold value as the characteristic point.
15. the threshold value waveform generating as described in any one of claim 1 to 14, it is characterised in that including noise processed
Portion, pre-treatment of the noise processed portion as the extraction process of the characteristic point in the feature extraction unit, for from described
Noise is removed in observation waveform.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2014094635A JP5791756B1 (en) | 2014-05-01 | 2014-05-01 | Threshold waveform generator |
| JP2014-094635 | 2014-05-01 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105022371A CN105022371A (en) | 2015-11-04 |
| CN105022371B true CN105022371B (en) | 2018-01-02 |
Family
ID=54346157
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410829332.0A Expired - Fee Related CN105022371B (en) | 2014-05-01 | 2014-12-22 | Threshold value waveform generating |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP5791756B1 (en) |
| KR (1) | KR101608155B1 (en) |
| CN (1) | CN105022371B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP6618227B1 (en) * | 2019-04-24 | 2019-12-11 | 有限会社 ソフトロックス | Data processing apparatus and data monitoring system |
| CN111950116A (en) * | 2020-06-23 | 2020-11-17 | 中铁第一勘察设计院集团有限公司 | Automatic simulation adjustment method for ballastless track of high-speed railway |
| CN112085391B (en) * | 2020-09-11 | 2024-02-27 | 山东钢铁集团日照有限公司 | Strip steel production process data characteristic value extraction method based on process rules |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2012151216A (en) * | 2011-01-18 | 2012-08-09 | Tokyo Electron Ltd | Processing apparatus abnormality determination system and abnormality determination method thereof |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3215722B2 (en) * | 1992-08-14 | 2001-10-09 | エヌイーシー三菱電機ビジュアルシステムズ株式会社 | Measurement waveform judgment method |
| JP3319076B2 (en) * | 1993-09-29 | 2002-08-26 | ソニー株式会社 | Waveform judgment device |
| JP4103029B2 (en) * | 2001-05-18 | 2008-06-18 | 有限会社 ソフトロックス | Process monitoring method |
| JP4239140B2 (en) | 2002-06-20 | 2009-03-18 | 有限会社 ソフトロックス | Data processing method using standard deviation |
| JP4104468B2 (en) * | 2003-02-10 | 2008-06-18 | 日置電機株式会社 | Waveform storage device and waveform determination area setting method |
| JP2006197033A (en) * | 2005-01-11 | 2006-07-27 | Sumitomo Electric Ind Ltd | Image photographing system |
| JP5030412B2 (en) * | 2005-10-27 | 2012-09-19 | 大和製罐株式会社 | Method and apparatus for determining whether or not welded portion of welded body is weldable |
| JP2008298525A (en) * | 2007-05-30 | 2008-12-11 | Mitsubishi Electric Engineering Co Ltd | Waveform abnormality detector |
| JP5205174B2 (en) * | 2008-08-08 | 2013-06-05 | 新明和工業株式会社 | Terminal crimping condition pass / fail judgment device, terminal crimping processing device |
| JP5353656B2 (en) * | 2009-11-24 | 2013-11-27 | 株式会社安川電機 | Robot controller |
| JP5419921B2 (en) | 2011-04-25 | 2014-02-19 | 三菱電機株式会社 | Inspection device |
-
2014
- 2014-05-01 JP JP2014094635A patent/JP5791756B1/en not_active Expired - Fee Related
- 2014-12-01 KR KR1020140169617A patent/KR101608155B1/en active IP Right Grant
- 2014-12-22 CN CN201410829332.0A patent/CN105022371B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2012151216A (en) * | 2011-01-18 | 2012-08-09 | Tokyo Electron Ltd | Processing apparatus abnormality determination system and abnormality determination method thereof |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2015212635A (en) | 2015-11-26 |
| KR20150126267A (en) | 2015-11-11 |
| CN105022371A (en) | 2015-11-04 |
| KR101608155B1 (en) | 2016-03-31 |
| JP5791756B1 (en) | 2015-10-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104545887B (en) | The recognition methods of artifact ecg wave form and device | |
| CN105022371B (en) | Threshold value waveform generating | |
| Chouhan et al. | Total removal of baseline drift from ECG signal | |
| CN101357076B (en) | Remote heart reservation index analysis system and method | |
| CN105468951B (en) | Method and device, the wearable device of identification are carried out by ecg characteristics | |
| CN110236573A (en) | The detection method and relevant apparatus of psychological pressure state | |
| CN102483442A (en) | Battery state estimation device and method of estimating battery state | |
| WO2005106534A3 (en) | Improving inter-subject coherence in dt-mri | |
| Saxena et al. | QRS detection using new wavelets | |
| CN105588577A (en) | Detection method and detection apparatus for abnormal step counting in exercise monitoring device | |
| CN107811631A (en) | Electrocardiosignal method for evaluating quality | |
| CN104732076A (en) | Method for extracting energy trace characteristic of side channel | |
| CN104698325B (en) | A kind of method of discrimination of the low-frequency oscillation of power system negative damping mechanism and forced oscillation | |
| CN109350031A (en) | A kind of photoplethysmographic signal feature point detecting method and device | |
| KR101510522B1 (en) | System for classification electrocardiogram signals | |
| CN102988041A (en) | Selective cardiac-magnetic signal averaging method in signal noise suppression | |
| CN101393237A (en) | Three phase on-line harmonic current real time monitoring system and working method thereof | |
| CN103027692B (en) | Dynamic spectrum data processing method based on uncertainty | |
| CN102509117A (en) | DTW (dynamic time warping) based method for automatically distinguishing abnormal measurements from plantar pressure measurement | |
| CN111724386B (en) | Data preprocessing method and system for multi-frequency imaging and multi-frequency imaging system | |
| CN107735018A (en) | Blood pressure than calculation procedure and stores the storage medium of the program than computing device, blood pressure than computational methods, blood pressure | |
| Sivaranjini et al. | Computationally-efficient pulse rate estimation from compressed ppg measurements for continuous vital signs monitoring | |
| CN112948755B (en) | Telemetry sine parameter interpretation method | |
| CN111130681B (en) | Method and device for determining noise transfer characteristics | |
| Sun et al. | A ECG waveform detection algorithm based on differential threshold and wavelet transform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180102 |