CN104020310B - A kind of motor yarn-feeding device method of detecting malfunction based on machine vision - Google Patents

Abstract

The invention provides a kind of motor yarn-feeding device method of detecting malfunction based on machine vision, PC outputs data to motor driver by I/O card, to drive motor yarn-feeding device, calculates the output valve expection rotating speed of each pattern wheel；PC triggers the realtime image data of image acquisition device motor yarn-feeding device by image pick-up card, calculates the actual speed of each pattern wheel；With output valve, the relatively actual speed of each pattern wheel expects whether rotating speed is consistent, if pattern wheel actual speed fluctuates allowed band beyond nominal error, then assert this motor yarn-feeding device remarkable action.The method that the present invention provides overcomes the deficiencies in the prior art, it is possible to Autonomous test also judges whether motor yarn-feeding device operating condition meets expection, it is simple to checks and gets rid of the fault of yarn-feeding device in rug-overtufting machine operating, improves and install the work efficiency safeguarded.

Description

A kind of motor yarn-feeding device method of detecting malfunction based on machine vision

Technical field

The present invention relates to the detection method of a kind of motor yarn-feeding device remarkable action based on machine vision, belong to and spin Fabric weaving technology field.

Background technology

Rug-overtufting machine is driven by Timing Belt by a small machine in motor yarn-feeding device jacquard weave combined mechanism One or two pattern wheel is constituted.Main shaft rotates a fixed angle (such as 360 degree) and completes a jacquard weave week Phase, drive motor speed to follow the tracks of the speed of mainshaft according to fixed proportion coefficient, thus drive pattern wheel to make yarn feeding amount Requirement is met a jacquard weave cycle.It is mounted with in a large number on the tufting machine of small machine yarn-feeding device jacquard weave owing to using Small machine yarn-feeding device, first install and operation exception after typically require hand inspection debug, tufting Artificial method for removing, the inefficiency one by one that on machine, small machine yarn feeding status checkout generally uses, usual one group carries Up to a hundred groups of the yarn-feeding device of flower, many thousands of groups, investigation is got up relatively difficult one by one, and working strength is bigger.

Summary of the invention

The technical problem to be solved in the present invention is to provide a kind of Aulomatizeted Detect motor yarn-feeding device action failure Method.

In order to solve above-mentioned technical problem, the technical scheme is that a kind of motor based on machine vision of offer Yarn-feeding device method of detecting malfunction, motor yarn-feeding device includes motor, and motor output end is by Timing Belt even Connecing the gear of pattern wheel side, pattern wheel is located on unpowered major axis, it is characterised in that: set on described pattern wheel There is the painting material strip of the enhancing contrast ratio of two auxiliary vision-based detection；PC drives dress by I/O card with motor Putting connection, motor driver connects described motor yarn-feeding device, and PC is adopted with image by image pick-up card Acquisition means connects；

The method specifically includes following steps:

Step 1:PC machine outputs data to motor driver, to drive motor yarn-feeding device by I/O card； Motor rotarily drives pattern wheel and rotates, if motor yarn-feeding device action is normal, then the rotating speed of pattern wheel should with work as The motor speed of front setting becomes fixed proportion relation, and motor speed follows main shaft according to a fixed proportion relation Rotating speed；Calculate the output valve expection rotating speed of each pattern wheel；

Step 2:PC machine triggers the real-time of image acquisition device motor yarn-feeding device by image pick-up card View data, calculates the actual speed of each pattern wheel；

Step 3: whether the actual speed comparing each pattern wheel is consistent with output valve expection rotating speed, if jacquard weave Wheel actual speed fluctuates allowed band beyond nominal error, then assert this motor yarn-feeding device remarkable action.

Preferably, in described step 1, the establishing method of I/O card output data is as follows:

The most often motor all actions in group full-jacquard mechanism, but every time except the motor yarn-feeding device of one group of full-jacquard mechanism Other group motor status outer are all consistent, altogether need to export q × b group data；Wherein q is full-jacquard mechanism sum, B is yarn feeding amount species number in single full-jacquard mechanism；

I & lt output data are defined as follows: [x_i1, x_i2, x_i3..., x_iq], wherein x_i1It is i-th group of data first Row ..., x_iqIt is i-th group of data q row；

For i-th group of data, i starts to q × b from 1,

Computation rule is started the cycle over as follows from the 1st row:

The first step: j=works as prostatitis sequence number, if (i-1)/b+1 ≠ j, current train value x_ij=b-(i-1) %b-1 > 0?B-(i-1) %b-1:b；If (i-1)/b+1==j, then current train value x_ij=b-(i-1) %b；% For the symbol that rems；

Second step: when prostatitis sequence number=when prostatitis sequence number+1, if when prostatitis sequence number > q, the then calculating of this group data Terminate, otherwise jump to the first step and continue executing with.

Preferably, in described step 1, the computational methods of the output valve expection rotating speed of pattern wheel are as follows:

Set any instant speed of mainshaft ω_main, set certain time period main shaft mean speedEach is defeated Go out data [1,2 ..., b] corresponding proportionality coefficient is [k₁, k₂..., k_b], b is single full-jacquard mechanism yarn feeding amount species number, Motor is k to pattern wheel gear ratio_{dj_thl}, random time section expection pattern wheel mean speed is ω_thl, any time Expection pattern wheel mean speed isOutput data j (j ∈ [and 1,2 ..., b]) period, corresponding proportionality coefficient For k_j, can obtain:

ω_{thl_qw}=ω_main×k_j×k_{dj_thl}

K in formula_{dj_thl}For the rotating speed proportionality coefficient of motor to pattern wheel, k_jFor exporting data j to current motor Time main shaft to the rotating speed proportionality coefficient of motor.

Encoder on main shaft is used to calculate the average speed in certain stage；Set and gather initial time encoder pulse Number n_start, gather end time encoder pulse number n_end, k_dFor each encoder pulse corresponding angle, during collection Between be T, then

$\stackrel{\‾}{{\mathrm{\ω}}_{\mathrm{thl}\_\mathrm{qw}}}=\stackrel{\‾}{{\mathrm{\ω}}_{\mathrm{main}}}\×k_j\×k_{\mathrm{dj}\_\mathrm{thl}}=k_j\×k_{\mathrm{dj}\_\mathrm{thl}}\×k_d\×(n_{\mathrm{end}}-n_{\mathrm{start}})/T$

The output valve expection mean speed of pattern wheel in arbitrary period can be obtained according to above formula.

Preferably, in described step 2, the computational methods of the actual speed of pattern wheel are as follows:

The painting material strip of 2 enhancing contrast ratio within one week, it is uniformly provided with, it is assumed that the collection of image collecting device pattern wheel Frequency is p frame/second, and on pattern wheel gear, enhancing contrast ratio is coated with material strip radius is R；

By gathering image, in order labelling image and acquisition time thereof, after Image semantic classification is corrected, intercept The image of area-of-interest, obtains enhancing contrast ratio by image background calculus of finite differences and is coated with material strip relative to gear rotation The position of axle；

y_n=R×cosθ_n, θ_n=arccos(y_n/ R)

y_n+m=R×cosθ_n+m, θ_n+m=arccos(y_n+m/ R)

Try to achieve actual measurement mean speed

$\stackrel{\‾}{{\mathrm{\ω}}_{\mathrm{thl}\_\mathrm{sc}}}=({\mathrm{\θ}}_{n+m}-{\mathrm{\θ}}_n)/t$

Wherein:

y_n、y_n+mRecord enhancing contrast ratio when being respectively n-th with the n-th+m time and be coated with material strip relative to axis height Degree；

θ_n、θ_n+mIt is respectively n-th and the angle rotated for the n-th+m time；

For currently recording mean angular velocity；

T is n-th and records image acquisition interval, t=m × 1/p for twice the n-th+m time；

It is derived from the actual average rotating speed of pattern wheel in arbitrary period.

Preferably, in described step 1, I/O card output data to ensure all each states of motor yarn-feeding device At least switching one time, and output data keep several jacquard weave cycles every time.

The method that the present invention provides overcomes the deficiencies in the prior art, it is possible to Autonomous test also judges motor yarn-feeding device Whether operating condition meets expection, it is simple to checks and gets rid of the fault of yarn-feeding device in rug-overtufting machine operating, improves The work efficiency safeguarded is installed.

Accompanying drawing explanation

Fig. 1 is detection apparatus system pie graph；

Fig. 2 is motor yarn-feeding device subsystem pie graph；

Fig. 3 is speed measuring and calculating coordinate schematic diagram；

The motor yarn-feeding device method of detecting malfunction workflow based on machine vision that Fig. 4 provides for the present invention Figure.

Detailed description of the invention

For making the present invention become apparent, hereby with a preferred embodiment, and accompanying drawing is coordinated to be described in detail below.

The method that the present invention provides controls the action of motor yarn-feeding device, electricity by software output signal specific combination After machine yarn-feeding device action, on meeting drive motor yarn-feeding device, pattern wheel rotates, and utilizes Vision Builder for Automated Inspection to gather electricity The work real-time video of machine yarn-feeding device, is analyzed video, it is determined that it is pre-whether pattern wheel spinning movement meets Phase.Then can quickly judge out of order motor yarn-feeding device position as do not met expection, the personnel of maintaining easily enter One step detection.

Rug-overtufting machine motor yarn-feeding device is generally made up of a motor and two intermeshing pattern wheels.As Shown in Fig. 1, this detection device by PC 1, I/O card 2, motor driver 3, image pick-up card 4, The inspection software 5 that is installed on PC, image collecting device 6, motor yarn-feeding device 7 form, PC 1 Being connected with motor driver 3 by I/O card 2, motor driver 3 connects motor yarn-feeding device 7, PC Being provided with inspection software 5 on machine, PC is connected with image collecting device 6 by image pick-up card 4.Image Harvester 6 the present embodiment uses camera.

In conjunction with Fig. 2, motor yarn-feeding device 7 is by motor 71, Timing Belt 72, unpowered major axis 73, pattern wheel 74, the painting material strip 75 of enhancing contrast ratio forms, and motor 71 outfan connects pattern wheel 74 by Timing Belt 72 The gear in left side, motor 71 rotates and is rotated by drive pattern wheel 74.Pattern wheel 74 is arranged on unpowered major axis On 73, pattern wheel 74 has the painting material strip 75 of the enhancing contrast ratio of two auxiliary vision-based detection.Strengthen contrast The painting material strip of degree is during camera is taken pictures, and the obvious color belt of background contrast.Such as pattern wheel is light, It is coated with material strip and just selects black.

Concrete detection method is implemented as follows:

All being driven by a motor 71 for each group of full-jacquard mechanism, under normal circumstances, if motor yarn feeding Mechanism action is normal, and drive mechanism is normal, then in jacquard attachment, pattern wheel and gear rotational speed thereon should be with The current motor speed set becomes fixed proportion relation, and motor speed follows master according to a fixed proportion relation Axle rotating speed.Utilize machine vision to obtain the real time imaging of jacquard attachment, calculate each pattern wheel by image procossing Rotating speed, fluctuate allowed band beyond nominal error if any rotating speed, then assert that this group full-jacquard mechanism has problems, Need the most manually to assert failure cause.

A) setting of data is exported: owing to motor speed is the most electrodeless adjustable, but real when being actually used in jacquard weave Border uses change limitednumber, within generally using 16 kinds of changes.Concrete output parameter establishing method has following Two kinds:

Method 1: every time only test a motor yarn-feeding device in all jacquard attachments, needs to export q × b altogether Group data.Wherein q is motor yarn-feeding device number, and b is motor yarn-feeding device yarn feeding amount change sum.

I & lt output data are defined as follows: [x_i1, x_i2, x_i3..., x_iq], wherein x_i1It is i-th group of data first Row ..., x_iqIt is i-th group of data q row.

For i-th group of data (i starts to q × b from 1),

Computation rule is started the cycle over as follows from the 1st row:

The first step: j=works as prostatitis sequence number, if (i-1)/b+1 ≠ j, current train value x_ijIt is 0；

If (i-1)/b+1==j, then current train value x_ij=b-(i-1) %b (in formula, % is the symbol that rems).

Second step: when prostatitis sequence number=when prostatitis sequence number+1, if when prostatitis sequence number > q, the then calculating of this group data Terminate, otherwise jump to the first step and continue executing with.

Q=5 as a example by Fig. 25 full-jacquard mechanisms, each motor only arranges three kinds of change b=3, exports data Q × b=5 × 3=15 group, method 1 sets as follows:

[3,0,0,0,0], [2,0,0,0,0], [1,0,0,0,0], [0,3,0,0,0], [0,2,0,0,0], [0,1, 0,0,0], [0,0,3,0,0], [0,0,2,0,0], [0,0,1,0,0], [0,0,0,3,0], [0,0,0,2,0], [0,0,0,1,0], [0,0,0,0,3], [0,0,0,0,2], [0,0,0,0,1].

Wherein 3 representing full-jacquard mechanism motor rotation at a high speed, 2 represent full-jacquard mechanism motor rotation in middling speed, and 1 Representing full-jacquard mechanism motor rotation at low speed, 0 represents this full-jacquard mechanism motor is failure to actuate.The often group every string of data Represent a full-jacquard mechanism output state.

Method 2: the most often motor all actions in group jacquard attachment, but every time except the motor of one group of jacquard attachment Yarn-feeding device other group motor status outer are all consistent, altogether need to export q × b group data.Wherein q is jacquard weave Mechanism's sum, b is yarn feeding amount species number in single full-jacquard mechanism.

I & lt output data are defined as follows: [x_i1, x_i2, x_i3..., x_iq], wherein x_i1It is i-th group of data first Row ..., x_iqIt is i-th group of data q row.

For i-th group of data (i starts to q × b from 1),

Computation rule is started the cycle over as follows from the 1st row:

The first step: j=works as prostatitis sequence number, if (i-1)/b+1 ≠ j, current train value x_ij=b-(i-1) %b-1 > 0?(in formula, % is the symbol that rems to b-(i-1) %b-1:b, if a%b is a divided by more than b Number, this expression formula is a kind of literary style in c language, if meaning is to set up for expressing before question mark, value question mark and Value between colon, otherwise takes colon value below)；

If (i-1)/b+1==j, then current train value x_ij=b-(i-1) %b (in formula, % is the symbol that rems).

Second step: when prostatitis sequence number=when prostatitis sequence number+1, if when prostatitis sequence number > q, the then calculating of this group data Terminate, otherwise jump to the first step and continue executing with.

Data setting is exported as follows as a example by Fig. 25 full-jacquard mechanisms:

Q=5 as a example by Fig. 25 full-jacquard mechanisms, each motor only arranges three kinds of yarn feeding amount change b=3, defeated Going out data q × b=5 × 3=15 group, it is as follows that method two tests data setting:

[3,2,2,2,2], [2,1,1,1,1], [1,3,3,3,3], [2,3,2,2,2], [1,2,1,1,1], [3,1, 3,3,3], [2,2,3,2,2], [1,1,2,1,1], [3,3,1,3,3], [2,2,2,3,2], [1,1,1,2,1], [3,3,3,1,3], [2,2,2,2,3], [1,1,1,1,2], [3,3,3,3,1].

Method 1 is relatively suitable for first installation and debugging, and method 2 all can use when first debugging and maintenance debugging, makees For preferably, the present invention uses method 2 to be set exporting data.

B) calculating of pattern wheel expection rotating speed: set any instant speed of mainshaft ω_main, set certain time period Main shaft mean speedEach output data [1,2 ..., b] corresponding proportionality coefficient is [k₁, k₂..., k_b] (b is Yarn feeding amount species number in single full-jacquard mechanism), motor to pattern wheel gear ratio is k_{dj_thl}, random time section is expected Pattern wheel mean speed is ω_{thl_qw}, arbitrary period expection pattern wheel mean speed isIn output data J (j ∈ [1,2 ..., b])) period, corresponding proportionality coefficient is k_jCan obtain:

ω_{thl_qw}=ω_main×k_j×k_{dj_thl}

K in formula_{dj_thl}For the rotating speed proportionality coefficient of motor to pattern wheel, k_jFor exporting data j to current motor Time main shaft to the rotating speed proportionality coefficient of motor.

But actually so measuring and calculating is relatively difficult, the present invention uses encoder on main shaft to calculate the flat of certain stage All speed.

Set and gather initial time encoder pulse number n_start, gather end time encoder pulse number n_end, often Individual encoder pulse corresponding angle k_d, acquisition time is T, then

$\stackrel{\‾}{{\mathrm{\ω}}_{\mathrm{thl}\_\mathrm{qw}}}=\stackrel{\‾}{{\mathrm{\ω}}_{\mathrm{main}}}\×k_j\×k_{\mathrm{dj}\_\mathrm{thl}}=k_j\×k_{\mathrm{dj}\_\mathrm{thl}}\×k_d\×(n_{\mathrm{end}}-n_{\mathrm{start}})/T$

The expectation mean speed of pattern wheel in arbitrary period can be obtained according to formula above.

C) the pattern wheel rotating speed computational methods of view-based access control model: process for convenience of successive image, pattern wheel one week It is uniformly provided with the painting material strip 75 of 2 enhancing contrast ratio.Assume that vision collecting system acquisition frequency is p frame/second, It is R that pattern wheel gear enhancing contrast ratio is coated with material strip radius.

By gathering image, in order labelling image and acquisition time thereof, after Image semantic classification is corrected, intercept The image of area-of-interest, obtains enhancing contrast ratio by image background calculus of finite differences and is coated with material strip relative to gear rotation The position of axle.As it is shown on figure 3, image schematic diagram when left side is n-th collection image, right side is the n-th+m Image schematic diagram during secondary collection image, understands y according to middle coordinate schematic diagram_n=R×cosθ_n, reverse θ_n=arccos(y_n/ R).

Same y_n+m=R×cosθ_n+m, then θ_n+m=arccos(y_n+m/ R).

Try to achieve actual measurement mean speed $\stackrel{\‾}{{\mathrm{\ω}}_{\mathrm{thl}\_\mathrm{sc}}}=({\mathrm{\θ}}_{n+m}-{\mathrm{\θ}}_n)/t.$

Wherein:

y_n、y_n+mRecord enhancing contrast ratio when being respectively n-th with the n-th+m time and be coated with material strip relative to axis height Degree；

θ_n、θ_n+mIt is respectively n-th and the angle rotated for the n-th+m time；

For currently recording mean angular velocity；

T be n-th and the n-th+m time two record image acquisition interval, t=m × 1/p；

Any one pattern wheel gear actual angular speed can be obtained, by comparing according to method described in above-mentioned formula This actual angular speed and corresponding jacquard attachment expection tachometer value, if speed error absolute valueLess than permissible value, then it is assumed that current measuring rotating speed meets the requirements, and is otherwise that rotating speed is different Normal or defective.

Specific implementation method step of the present invention is as shown in Figure 4:

Step 1: setup parameter, sets the speed of mainshaft, sets output data, and every speed parameter is relative In speed of mainshaft proportionality coefficient, pattern wheel radius and image acquisition interval.Wherein output data to ensure All each states of motor yarn-feeding device at least switch one time, the number of examples in output data such as said method 2 According to.

Step 2: export i-th group of data by I/O card, drives motor yarn-feeding device, ensures several as required Individual jacquard weave cycle data is consistent, facilitates subsequent image acquisition to process.

Step 3: trigger collected by camera full-jacquard mechanism realtime image data by image pick-up card and calculate each and carry Floral whorl rotating speed.As preferably, in use collection image, the painting material strip of enhancing contrast ratio is near position, pattern wheel axis The image put carries out speed calculation, uses y in this example_n∈ [-R/2.0, R/2.0], y_n+m∈ [-R/2.0, R/2.0], and | y_n+m-y_n| ＞ R/2.0 image measuring and calculating mean speed is to reduce error.

Step 4: compare whether each pattern wheel rotating speed is consistent with output valve expection rotating speed, and export this detection Result.

Step 5: detected whether, as completed, skips to step 7.

Step 6: current group number i=i+1, skips to step 2 and performs.

Step 7: detection of end.

Claims (4)

1. a detection method for motor yarn-feeding device remarkable action based on machine vision, motor yarn-feeding device (7) Including motor (71), motor (71) outfan connects pattern wheel (74) side by Timing Belt (72) Gear, pattern wheel (74) is located on unpowered major axis (73), it is characterised in that: described pattern wheel (74) It is provided with the painting material strip (75) of the enhancing contrast ratio of two auxiliary vision-based detection；PC (1) passes through I/O Card (2) is connected with motor driver (3), and motor driver (3) connects described motor yarn-feeding device (7), PC (1) is connected with image collecting device (6) by image pick-up card (4)；

The method specifically includes following steps:

Step 1:PC machine (1) outputs data to motor driver (3) by I/O card (2), to drive Motor yarn-feeding device (7)；

Motor (71) rotarily drives pattern wheel (74) and rotates, if motor yarn-feeding device (7) action is normal, Then the rotating speed of pattern wheel (74) should become fixed proportion relation with the current motor speed set, and motor speed is pressed The speed of mainshaft is followed according to a fixed proportion relation；Calculate the output valve expection rotating speed of each pattern wheel (74)；

Step 2:PC machine (1) triggers image collecting device (6) by image pick-up card (4) and gathers motor The realtime image data of yarn-feeding device (7), calculates the actual speed of each pattern wheel (74)；

Step 3: whether the actual speed comparing each pattern wheel (74) is consistent with output valve expection rotating speed, as Really pattern wheel (74) actual speed fluctuates allowed band beyond nominal error, then assert this motor yarn-feeding device (7) remarkable action；

Wherein, in described step 2, the computational methods of the actual speed of pattern wheel are as follows:

The painting material strip (75) of 2 enhancing contrast ratio within one week, it is uniformly provided with, it is assumed that image collecting device pattern wheel (6) frequency acquisition is p frame/second, and on pattern wheel gear, enhancing contrast ratio is coated with material strip radius is R；

By gathering image, in order labelling image and acquisition time thereof, after Image semantic classification is corrected, intercept The image of area-of-interest, obtains enhancing contrast ratio by image background calculus of finite differences and is coated with material strip relative to gear rotation The position of axle；

y_n=R × cos θ_n, θ_n=arccos (y_n/R)

y_n+m=R × cos θ_n+m, θ_n+m=arccos (y_n+m/R)

Try to achieve

$\stackrel{\‾}{{\mathrm{\ω}}_{thl\_sc}}=({\mathrm{\θ}}_{n+m}-{\mathrm{\θ}}_n)/t$

Wherein:

y_n、y_n+mRecord enhancing contrast ratio when being respectively n-th with the n-th+m time and be coated with material strip relative to axis height Degree；

θ_n、θ_n+mIt is respectively n-th and the angle rotated for the n-th+m time；

For currently recording mean angular velocity；

T is n-th and records image acquisition interval, t=m × 1/p for twice the n-th+m time.

A kind of detection method of motor yarn-feeding device remarkable action based on machine vision, It is characterized in that: in described step 1, the establishing method of I/O card (2) output data is as follows:

Every time the most often motor all actions in group motor yarn-feeding device, but every time in addition to one group of motor yarn-feeding device other The motor status of group motor yarn-feeding device is all consistent, altogether needs to export q × b group data；Wherein q is motor Yarn-feeding device sum, b is yarn feeding amount species number in single motor yarn-feeding device；

I & lt output data are defined as follows: [x_i1,x_i2,x_i3,...,x_iq], wherein x_i1It is i-th group of data first Row ..., x_iqIt is i-th group of data q row；

For i-th group of data, i starts to q × b from 1,

Computation rule is started the cycle over as follows from the 1st row:

The first step: j=works as prostatitis sequence number, if (i-1)/b+1 ≠ j, current train value x_ij =b-(i-1) %b-1 > 0？B-(i-1) %b-1:b；If (i-1)/b+1==j, then current train value x_ij=b-(i-1) %b；% is Rem symbol；

Second step: when prostatitis sequence number=when prostatitis sequence number+1, if when prostatitis sequence number > q, the then calculating of this group data Terminate, otherwise jump to the first step and continue executing with.

A kind of detection method of motor yarn-feeding device remarkable action based on machine vision, It is characterized in that: in described step 1, the computational methods of the output valve expection rotating speed of pattern wheel are as follows:

Set any instant speed of mainshaft ω_main, set certain time period main shaft mean speedEach is defeated Go out data [1,2 ..., b] corresponding proportionality coefficient is [k₁,k₂,...,k_b], motor to pattern wheel gear ratio is k_{dj_thl}, appoint Meaning time period expection pattern wheel mean speed is ω_{thl_qw}, any time expection pattern wheel mean speed is k_jFor to current motor output data j time main shaft to the rotating speed proportionality coefficient of motor, wherein j ∈ [1,2 ..., b], Can obtain:

ω_{thl_qw}=ω_main×k_j×k_{dj_thl}

Encoder on main shaft is used to calculate the average speed in certain stage；Set and gather initial time encoder pulse Number n_start, gather end time encoder pulse number n_end, k_dFor each encoder pulse corresponding angle, during collection Between be T, then

$\stackrel{\‾}{{\mathrm{\ω}}_{thl\_qw}}=\stackrel{\‾}{{\mathrm{\ω}}_{main}}\×k_j\×k_{dj\_thl}=k_j\×k_{dj\_thl}\×k_d\×(n_{end}-n_{start})/T$

Any time expection pattern wheel mean speed can be obtained according to above formula.

A kind of detection method of motor yarn-feeding device remarkable action based on machine vision, It is characterized in that: in described step 1, I/O card (2) output data to ensure all motor yarn-feeding devices each State at least switches one time, and output data keep several jacquard weave cycles every time.