CN113378462A - Self-learning laser power fluctuation identification method and system - Google Patents

Self-learning laser power fluctuation identification method and system Download PDF

Info

Publication number
CN113378462A
CN113378462A CN202110642357.XA CN202110642357A CN113378462A CN 113378462 A CN113378462 A CN 113378462A CN 202110642357 A CN202110642357 A CN 202110642357A CN 113378462 A CN113378462 A CN 113378462A
Authority
CN
China
Prior art keywords
power
laser
self
learning
repetition frequency
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110642357.XA
Other languages
Chinese (zh)
Other versions
CN113378462B (en
Inventor
徐永顺
王铁男
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin Kaipulin Laser Technology Co ltd
Original Assignee
Tianjin Kaipulin Laser Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tianjin Kaipulin Laser Technology Co ltd filed Critical Tianjin Kaipulin Laser Technology Co ltd
Priority to CN202110642357.XA priority Critical patent/CN113378462B/en
Publication of CN113378462A publication Critical patent/CN113378462A/en
Application granted granted Critical
Publication of CN113378462B publication Critical patent/CN113378462B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J1/00—Photometry, e.g. photographic exposure meter
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • Lasers (AREA)

Abstract

The application provides a method and a system for recognizing self-learning laser power fluctuation, wherein the method for recognizing the self-learning laser power fluctuation comprises the following steps: s100: and (3) self-learning stage: acquisition of laser Aij(i is 1 … n, j is 1 … m), i is the number of pulse trains, and j is the repetition frequency; obtaining each laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij) (ii) a All average powers P (A)ij) Fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power; s200: a detection stage: and acquiring the actual power and the reference power of the laser to be detected, and calculating a fluctuation difference value. The method for recognizing the self-learning laser power fluctuation can recognize the power fluctuation, ensure that the problem can be found at the first time, and reduceAnd (4) economic loss.

Description

Self-learning laser power fluctuation identification method and system
Technical Field
The present disclosure relates generally to the field of laser power detection technologies, and in particular, to a method and a system for self-learning laser power fluctuation identification.
Background
The ultrafast laser is a laser with short laser pulse width less than 15 picoseconds, high single pulse energy and high repetition frequency and has good processing effect on brittle materials, and the laser is usually a picosecond laser and a femtosecond laser in the industry. The repetition frequency of the laser is usually 1kHz-1000KHz selectable, the number of pulse trains is adjustable from 1 to 10, and different repetition frequencies and the number of pulse trains are applied to different processing technologies.
When the laser is normally used, a client can change the repetition frequency and the number of pulse strings of the laser when processing different materials, the repetition frequency and the number of pulse penetration are changed, and the normal output power of the laser can also be changed.
In the practical use process, when the power fluctuation is overlarge, the rear end processing is easy to fail, the normal production progress is influenced, and even economic loss is generated. Therefore, a method or system for identifying laser power fluctuation is needed to ensure that the problem can be found in the first time and the economic loss is reduced.
Disclosure of Invention
In view of the above-mentioned drawbacks and deficiencies of the prior art, it would be desirable to provide a method and system for self-learning laser power fluctuation identification,
in a first aspect, the present application provides a method for self-learning laser power fluctuation identification, comprising the following steps:
s100: and (3) self-learning stage:
acquisition of laser Aij(i is 1 … n, j is 1 … m), i is the number of pulse trains, and j is the repetition frequency;
detecting each of the lasers A a multiple timesij(i-1 … n, j-1 … m) and obtaining the output power of each laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij)；
All the average powers P (A)ij) Fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power;
s200: a detection stage:
and acquiring the actual power and the reference power of the laser to be detected, and calculating a fluctuation difference value.
According to the technical scheme provided by the embodiment of the application, each laser A is obtainedij(i-1 … n, j-1 … m)Average power P (A)ij) The method specifically comprises the following steps:
for each laser AijDetecting the output power of (i-1 … n, j-1 … m) for multiple times to obtain multiple data samples;
removing a maximum value and a minimum value from a plurality of the data samples;
obtaining a mean of the remaining data samples as the average power P (A)ij)。
According to the technical scheme provided by the embodiment of the application, the method for forming the binary piecewise function prediction model specifically comprises the following steps:
average power P (A) with the same number i of pulse trainsij) Dividing into a group;
for all average powers P (A) in each groupij) Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power;
all the function units form the binary piecewise function prediction model.
According to the technical scheme provided by the embodiment of the application, the forming of the function unit specifically comprises:
setting the functional relation between the reference power and the repetition frequency j as follows:
P＝f(j)＝a*lnj+b
define the sum of the squares of the total errors:
Figure BDA0003107537280000021
adding Se2Respectively solving partial derivatives of a and b;
adding Se2Taking the minimum value and associating each repetition frequency j with its corresponding average power P (A)ij) The values of a and b are obtained and are put into the functional relation.
According to the technical scheme provided by the embodiment of the application, the method further comprises the following steps after the fluctuation difference value is calculated:
and sending alarm information when the fluctuation difference value is judged to be larger than a set value.
In a second aspect, the present application provides a self-learning laser power fluctuation identification system, comprising:
laser for emitting laser light Aij(i＝1…n，j＝1…m)；
The control module is connected with the laser and is used for controlling the number i of the laser pulse trains emitted by the laser and the repetition frequency j;
a power detection module for detecting the laser Aij(i-1 … n, j-1 … m);
a power self-learning module: for calculating each of the laser lights AijAverage power P (a) of (i 1 … n, j 1 … m)ij) And all the average powers P (A) are calculatedij) Fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power;
and the power prediction module is used for outputting the reference power value of the emitted laser through the binary piecewise function prediction model.
According to the technical scheme provided by the embodiment of the application, the power self-learning module comprises: the device comprises a power data acquisition unit, a power data statistical unit and a power self-learning unit;
the power data acquisition unit is connected with the control module and the power detection module; the power data acquisition unit is configured to: sending a control instruction to the control module to change the laser AijThe number i of bursts and the repetition frequency j of (i 1 … n, j 1 … m); obtaining each laser A detected by the power detection moduleij(i-1 … n, j-1 … m);
the power data statistical unit is connected with the power data acquisition unit and used for calculating each laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij)；
The power self-learning unit is connected with the power data statistical unit and is used for calculating all the average power P (A)ij) Fitting to form a binary piecewise function prediction model with independent variables of pulse train number i and repetition frequency j and function value of reference power。
According to the technical scheme provided by the embodiment of the application, the power self-learning unit is specifically configured to:
average power P (A) with the same number i of pulse trainsij) Dividing into a group;
for all average powers P (A) in each groupij) Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power;
all the function units form the binary piecewise function prediction model.
According to the technical scheme provided by the embodiment of the application, the forming of the function unit specifically comprises:
setting the functional relation between the reference power and the repetition frequency j as follows:
P＝f(j)＝a*lnj+b
define the sum of the squares of the total errors:
Figure BDA0003107537280000041
adding Se2Respectively solving partial derivatives of a and b;
adding Se2Taking the minimum value and associating each repetition frequency j with its corresponding average power P (A)ij) And (4) solving a and b and putting the values into the functional relation.
According to the technical scheme provided by the embodiment of the application, the power alarm device further comprises a power alarm module; the power alarm module comprises a comparison unit and an alarm unit;
the comparison unit is configured to obtain the actual power and the reference power of the laser to be detected and calculate a fluctuation difference value; and the alarm unit is configured to send alarm information when the fluctuation difference value is judged to be larger than a set value.
The beneficial effect of this application lies in: by obtaining laser A with different pulse train numbers i and repetition frequencies jij(i-1 … n, j-1 … m) and detecting the average power value, and fitting all the average powers and the corresponding pulse train numbers i and repetition frequencies j to form independent variables of the pulse train numbers i and jA binary piecewise function prediction model with the repetition frequency j and the function value as reference power;
the binary piecewise function prediction model is obtained, so that the reference power can be predicted for the laser emitted by different pulse string numbers i and repetition frequencies j, and the laser fluctuation of different pulse string numbers i and repetition frequencies j is recognized by taking the corresponding reference power as a reference quantity;
in the process of the detection stage, the number i of the pulse trains of the laser to be detected and the repetition frequency j are input into the binary piecewise function prediction model, and the reference power under the number i of the pulse trains and the repetition frequency j can be obtained; and finally, comparing the reference power with the actual power to calculate a fluctuation difference value.
Through obtaining reference power and actual power for the fluctuation difference can quantify, in the actual production process of being convenient for, but problem is found to the very first time when the fluctuation is unusual, reduces economic loss.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings in which:
fig. 1 is a flowchart of a method for self-learning laser power fluctuation identification provided by the present application.
FIG. 2 is a flowchart of the method of step S120 shown in FIG. 1;
fig. 3 is a self-learning laser power fluctuation identification system provided by the present application.
Reference numbers in the figures: 1. a laser; 2. a control module; 3. a power detection module; 4. a power self-learning module; 5. a power prediction module; 6. a power data acquisition unit; 7. a power data statistics unit; 8. a power self-learning unit; 9. a power alarm module; 10. a comparison unit; 11. and an alarm unit.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
Example 1
Please refer to fig. 1, which is a method for self-learning laser power fluctuation recognition provided by the present application:
s100: and (3) self-learning stage:
s110: acquisition of laser Aij(i is 1 … n, j is 1 … m), i is the number of pulse trains, and j is the repetition frequency;
specifically, the number of pulse trains and repetition frequency are physical quantities representing laser, the number of pulse trains of the laser is 1-10 in a normal state, and the repetition frequency of the laser is 100Hz-1000 Hz; therefore, for convenience of describing the technical solution of the present application, in the present embodiment, n is 10, and m is 19, which can be set by other workers in the field according to actual needs.
i＝1 i＝2 i＝3 …… i＝10
Number of pulse trains 1 2 3 10
TABLE-1
j＝1 j＝2 j＝3 …… j＝19
Repetition frequency 100 150 200 …… 1000
TABLE-2
As shown in tables-1 and 2, the number of bursts is 10, and the number of repetition frequencies is 19; thus the laser Aij(i-1 … n, j-1 … m) from A11To A1019And the number of the grooves is 190.
S120: obtain each instituteThe laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij)；
Specifically, the average power P (A) of each laser is obtainedij) By measuring laser A multiple times11And obtaining the average value through an algorithm to obtain the average power P (A)ij)。
S130: all the average powers P (A)ij) Fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power;
s200: a detection stage:
s210: acquiring the actual power and the reference power of the laser to be detected;
and when the reference power of the laser to be detected is obtained, the repetition frequency and the number of the pulse strings of the laser to be detected are input into the binary piecewise function prediction model, and the reference power of the laser to be detected is calculated.
The actual power of the laser to be detected can be obtained and detected through the detection device.
S220: and calculating a fluctuation difference value.
And calculating a fluctuation difference value, namely subtracting the reference power from the actual power to obtain the fluctuation difference value. And comparing the obtained fluctuation difference value with a set threshold value, and outputting alarm information when the fluctuation difference value is larger than the set threshold value.
The working principle is as follows: the self-learning stage is mainly used for obtaining a binary piecewise function prediction model, namely, the laser A with different pulse train numbers i and repetition frequencies j is obtainedij(i is 1 … n, j is 1 … m) and detecting the average power value thereof, and fitting all the average powers and the corresponding pulse train numbers i and repetition frequencies j to form a binary piecewise function prediction model with independent variables of the pulse train numbers i and the repetition frequencies j and a function value of the binary piecewise function prediction model as reference power;
in the process of the detection stage, the number i of the pulse trains of the laser to be detected and the repetition frequency j are input into the binary piecewise function prediction model, and the reference power under the number i of the pulse trains and the repetition frequency j can be obtained; and finally, comparing the reference power with the actual power to calculate a fluctuation difference value. In the above step, the reference power can be predicted for the laser emitted by different numbers i of pulse trains and repetition frequencies j by obtaining the binary piecewise function prediction model; through obtaining reference power and actual power for the fluctuation difference can quantify, the very first time discovery problem of being convenient for reduces economic loss.
Wherein, in a preferred embodiment, as shown in fig. 2, each of the laser lights a is acquiredijAverage power P (a) of (i 1 … n, j 1 … m)ij) The method specifically comprises the following steps:
s121: for each laser AijDetecting the output power of (i-1 … n, j-1 … m) for multiple times to obtain multiple data samples;
specifically, each of the lasers AijDetecting the output power of (i-1 … n, j-1 … m) for 600 times to obtain 600 data samples; the total number of samples was 600 × 190 — 114000.
It should be noted that the detection times can be set according to actual requirements, and the detection times are set to 600 times in the embodiment, so that the accuracy of the detection result is improved; the actual process can reduce the number of detection times under the condition of ensuring the accuracy of the test result so as to improve the detection speed, and for example, the number can be set to 20.
S122: removing a maximum value and a minimum value from a plurality of the data samples;
s123: obtaining a mean of the remaining data samples as the average power P (A)ij)。
Specifically, after the maximum value and the minimum value are removed from 600 data samples corresponding to each laser, 598 data are remained, and the average power P (a) can be obtained by performing average calculation on the 598 dataij). By removing the maximum value and the minimum value, abnormal data or error data generated by measurement or external factors are reduced, so that the finally obtained average value is more accurate.
In a preferred embodiment, the method for forming the binary piecewise function prediction model specifically includes: average power P (A) with the same number i of pulse trainsij) Dividing into a group; with i equal to 1 and m equal toFor example, 19 is shown in the following table:
serial number Laser Aij Number of pulse trains Repetition frequency (HZ) P(Aij)(0.01w)
1 Laser A11 1 100 1627
2 Laser A12 1 150 1684
3 Laser A13 1 200 1717
4 Laser A14 1 250 1753
5 Laser A15 1 300 1767
6 Laser A16 1 350 1783
7 Laser A17 1 400 1800
8 Laser A18 1 450 1814
9 Laser A19 1 500 1826
10 Laser A110 1 550 1837
11 Laser A111 1 600 1848
12 Laser A112 1 650 1857
13 Laser A113 1 700 1867
14 Laser A114 1 750 1877
15 Laser A115 1 800 1886
16 Laser A116 1 850 1892
17 Laser A117 1 900 1898
18 Laser A118 1 950 1905
19 Laser A119 1 1000 1915
TABLE-3
For all average powers P (A) in each groupij) Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power;
all the function units form the binary piecewise function prediction model.
Specifically, m average powers P (A) having the same number i of bursts are setij) The method comprises the following steps of (1) dividing the raw materials into a group of n groups; m average powers P (A) in each groupij) And its corresponding repetition frequency j
Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power; and forming the binary piecewise function prediction model by the n function units.
In a preferred embodiment, the forming the function unit specifically includes:
setting the functional relation between the reference power P and the repetition frequency j as follows:
P＝f(j)＝a*lnj+b
in the formula, a and b are undetermined coefficients;
to make the function value of the obtained functional relation closer to the actual value, the sum of the squares of the total error is defined:
Figure BDA0003107537280000081
since different a and b will obtain different Se2According to multivariate calculus, Se2And (3) respectively calculating partial derivatives of a and b:
Figure BDA0003107537280000082
Figure BDA0003107537280000083
adding Se2Taking the minimum value, i.e. Se2When 0, the finishing can give:
a*2*[lnj1+lnj2+......+lnj19]+b*2*(lnj1+lnj2+lnj19)＝2(P1*lnj1+P2*lnj2+......+P19*lnj19)
a*2*(lnj1+lnj2+......+lnj19)+b*2*19＝2*(P1+P2+......+P19)
each repetition frequency j is associated with its corresponding average power P (A)ij) The values of a and b are obtained and are put into the functional relation.
Finally, obtaining a binary piecewise function prediction model:
Figure BDA0003107537280000091
to facilitate the explanation of the technical solutions provided in the present application, take i ═ 1 as an example and obtain ping through table-3Average power P (A)ij) And its corresponding repetition frequency j:
i.e. mixing P1＝1672(0.01W)，j1＝100(HZ)；P2＝1684(0.01W)，j2＝150(HZ)；……；P19＝1915(0.01W)，j19Substitution into the above formula for 1000 (HZ);
and obtaining the values of a and b, wherein the obtained result is as follows: 121.2 for a, 1074.4 for b;
the values of a and b are brought into a function relation of the set reference power P and the repetition frequency j to obtain a function relation
P＝121.2*lnj+1074.4 i＝1
Similarly, when i is 2, i is 3, … …, and i is m, the functional relationship between the reference power P and the repetition frequency j can be obtained by the above method.
It can be appreciated that the present application is implemented by applying an average power P (A) having the same number i of burstsij) Dividing into a group; the average power P (A) with the same repetition frequency j can also be usedij) The algorithm process is the same as the above process in principle, and thus is not described in detail.
In a preferred embodiment, the following steps are further included after calculating the fluctuation difference value:
and sending alarm information when the fluctuation difference value is judged to be larger than a set value. Through this step for the staff can acquire alarm information the very first time, avoids producing the production loss.
Example 2
The present application further provides a self-learning laser power fluctuation recognition system, as shown in fig. 2, including:
laser 1 for emitting laser light Aij(i＝1…n，j＝1…m)；
The control module 2 is connected with the laser and is used for controlling the number i of laser pulse trains emitted by the laser and the repetition frequency j;
a power detection module 3 for detecting the laser Aij(i-1 … n, j-1 … m);
the power self-learning module 4: for calculating each of the laser lights AijAverage power P (a) of (i 1 … n, j 1 … m)ij) And all the average powers P (A) are calculatedij) Fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power;
and the power prediction module 5 is used for outputting the reference power value of the emitted laser through the binary piecewise function prediction model.
The working principle is as follows: the control module 2 controls the laser 1 to change the number i of pulse trains and the repetition frequency j, the power detection module 3 detects the output power of the laser with different numbers i of pulse trains and repetition frequencies j and sends the output power to the power self-learning module 4, and the power self-learning module 4 calculates the average power P (A) of the laser with different numbers i of pulse trains and repetition frequencies j through the obtained output power dataij) While simultaneously applying the average power P (A)ij) And fitting the pulse strings i and the repetition frequencies j corresponding to the pulse strings to form a binary piecewise function prediction model with the function value as the reference power.
After the self-learning stage is finished, (namely after a binary piecewise function prediction model is generated), the laser 1 emits laser to be detected, and the power detection module 3 detects the laser to be detected so as to obtain the actual power of the real-time laser; the pulse train number i and the repetition frequency j of the laser to be detected are input into the binary piecewise function prediction model, so that the reference power under the pulse train number i and the repetition frequency j can be obtained, and the fluctuation difference value can be calculated by comparing the reference power with the actual power.
The self-learning laser power fluctuation identification system is simple in structure, a binary piecewise function prediction model can be formed through self-learning fitting, reference power under different pulse train numbers i and repetition frequencies j can be accurately obtained, meanwhile, with the help of the reference power, fluctuation values can be quantized, fluctuation abnormity can be found at the first time, and economic loss is reduced.
Wherein, in a preferred embodiment of the power self-learning module, the power self-learning module comprises: the device comprises a power data acquisition unit 6, a power data statistical unit 7 and a power self-learning unit 8;
the power data acquisition unit 6 is connected with the control module 2 and the power detection module 3; the power data acquisition unit 6 is configured to: sending a control instruction to the control module 2 to change the laser AijThe number i of bursts and the repetition frequency j of (i 1 … n, j 1 … m); obtaining each laser A detected by the power detection module 3ij(i-1 … n, j-1 … m);
the power data statistical unit 6 is connected with the power data acquisition unit 7 and is used for calculating each laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij)；
The power self-learning unit 8 is connected with the power data statistical unit 7 and is used for calculating all the average power P (A)ij) And fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power.
In a preferred embodiment of the power self-learning unit 7, the power self-learning unit 7 is specifically configured to:
average power P (A) with the same number i of pulse trainsij) Dividing into a group;
for all average powers P (A) in each groupij) Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power;
all the function units form the binary piecewise function prediction model.
In a preferred embodiment, the forming the function unit specifically includes:
setting the functional relation between the reference power and the repetition frequency j as follows:
P＝f(j)＝a*lnj+b
define the sum of the squares of the total errors:
Figure BDA0003107537280000111
adding Se2Respectively solving partial derivatives of a and b;
adding Se2Taking the minimum value and associating each repetition frequency j with its corresponding average power P (A)ij) And (4) solving a and b and putting the values into the functional relation.
The process is as follows:
Figure BDA0003107537280000112
Figure BDA0003107537280000113
adding Se2Taking the minimum value, i.e. Se2When 0, the finishing can give:
a*2*[lnj1+lnj2+......+lnj19]+b*2*(lnj1+lnj2+lnj19)＝2(P1*lnj1+P2*lnj2+......+P19*lnj19)
a*2*(lnj1+lnj2+......+lnj19)+b*2*19＝2*(P1+P2+......+P19)
each repetition frequency j is associated with its corresponding average power P (A)ij) The values of a and b are obtained and are put into the functional relation.
Finally, obtaining a binary piecewise function prediction model:
Figure BDA0003107537280000121
for convenience of explanation of the technical solutions provided in the present application, i ═ 1 is taken as an example, and the average power P (a) is obtained through the above table-3ij) And its corresponding repetition frequency j:
i.e. mixing P1＝1672(0.01W)，j1＝100(HZ)；P2＝1684(0.01W)，j2＝150(HZ)；……；P19＝1915(0.01W)，j19Substitution into the above formula for 1000 (HZ);
and obtaining the values of a and b, wherein the obtained result is as follows: 121.2 for a, 1074.4 for b;
the values of a and b are brought into a function relation of the set reference power P and the repetition frequency j to obtain a function relation
P＝121.2*lnj+1074.4 i＝1
Similarly, when i is 2, i is 3, … …, and i is m, the functional relationship between the reference power P and the repetition frequency j can be obtained by the above method.
It can be appreciated that the present application is implemented by applying an average power P (A) having the same number i of burstsij) Dividing into a group; the average power P (A) with the same repetition frequency j can also be usedij) The algorithm process is the same as the above process in principle, and thus is not described in detail.
In a preferred embodiment, the self-learning laser power fluctuation identification system further comprises a power alarm module 9; the power alarm module 9 comprises a comparison unit 10 and an alarm unit 11;
the comparison unit 10 is configured to obtain the actual power and the reference power of the laser to be measured, and calculate a fluctuation difference; the alarm unit 11 is configured to send alarm information when the fluctuation difference is greater than a set value. Through setting up alarm module 9 for the staff can acquire alarm information the very first time, avoids producing the production loss.
The above description is only a preferred embodiment of the application and is illustrative of the principles of the technology employed. It will be appreciated by a person skilled in the art that the scope of the invention as referred to in the present application is not limited to the embodiments with a specific combination of the above-mentioned features, but also covers other embodiments with any combination of the above-mentioned features or their equivalents without departing from the inventive concept. For example, the above features may be replaced with (but not limited to) features having similar functions disclosed in the present application.

Claims (10)

1. A method for self-learning laser power fluctuation recognition is characterized by comprising the following steps: the method comprises the following steps:
s100: and (3) self-learning stage:
acquisition of laser Aij(i is 1 … n, j is 1 … m), i is the number of pulse trains, and j is the repetition frequency;
detecting each of the lasers A a multiple timesij(i-1 … n, j-1 … m) and obtaining the output power of each laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij)；
All the average powers P (A)ij) Fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power;
s200: a detection stage:
and acquiring the actual power and the reference power of the laser to be detected, and calculating a fluctuation difference value.
2. The method for self-learning laser power fluctuation recognition according to claim 1, wherein: obtaining each of the lasers AijAverage power P (a) of (i 1 … n, j 1 … m)ij) The method specifically comprises the following steps:
for each laser AijDetecting the output power of (i-1 … n, j-1 … m) for multiple times to obtain multiple data samples;
removing a maximum value and a minimum value from a plurality of the data samples;
obtaining a mean of the remaining data samples as the average power P (A)ij)。
3. The method for self-learning laser power fluctuation recognition according to claim 1, wherein: the method for forming the binary piecewise function prediction model specifically comprises the following steps:
average power P (A) with the same number i of pulse trainsij) Dividing into a group;
for all average powers P (A) in each groupij) Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power;
all the function units form the binary piecewise function prediction model.
4. A self-learning laser power fluctuation identification as claimed in claim 3, wherein: the forming of the function unit is specifically:
setting the functional relation between the reference power and the repetition frequency j as follows:
P＝f(j)＝a*lnj+b
define the sum of the squares of the total errors:
Figure FDA0003107537270000021
adding Se2Respectively solving partial derivatives of a and b;
adding Se2Taking the minimum value and associating each repetition frequency j with its corresponding average power P (A)ij) The values of a and b are obtained and are put into the functional relation.
5. A method of self-learning laser power fluctuation identification as claimed in any one of claims 1 to 4, wherein: after the fluctuation difference value is calculated, the method further comprises the following steps:
and sending alarm information when the fluctuation difference value is judged to be larger than a set value.
6. A self-learning laser power fluctuation identification system is characterized in that: the method comprises the following steps:
a laser (1) for emitting laser light Aij(i＝1…n，j＝1…m)；
The control module (2) is connected with the laser and is used for controlling the number i of laser pulse trains emitted by the laser and the repetition frequency j;
a power detection module (3) for detecting the laser Aij(i-1 … n, j-1 … m);
power self-learning module (4): for calculating each of the laser lights AijAverage power P (a) of (i 1 … n, j 1 … m)ij) And all the average powers P (A) are calculatedij) Fitting to form independent variables of the number i of the pulse trains and the repetition frequency j, functionA binary piecewise function prediction model with the numerical value as the reference power;
and the power prediction module (5) is used for outputting the reference power value of the emitted laser through the binary piecewise function prediction model.
7. The self-learning laser power fluctuation identification system according to claim 6, wherein: the power self-learning module comprises: the device comprises a power data acquisition unit (6), a power data statistical unit (7) and a power self-learning unit (8);
the power data acquisition unit (6) is connected with the control module (2) and the power detection module (3); the power data acquisition unit (6) is configured to: sending a control instruction to the control module (2) to change the laser AijThe number i of bursts and the repetition frequency j of (i 1 … n, j 1 … m); obtaining each laser A detected by the power detection module (3)ij(i-1 … n, j-1 … m);
the power data statistical unit (6) is connected with the power data acquisition unit (7) and is used for calculating each laser AijAverage power P (a) of (i 1 … n, j 1 … m)ij)；
The power self-learning unit (8) is connected with the power data statistical unit (7) and is used for transmitting all the average power P (A)ij) And fitting to form a binary piecewise function prediction model with independent variables of the number i of the pulse trains and the repetition frequency j and function values of the reference power.
8. The self-learning laser power fluctuation identification system according to claim 7, wherein: the power self-learning unit (7) is specifically configured to:
average power P (A) with the same number i of pulse trainsij) Dividing into a group;
for all average powers P (A) in each groupij) Fitting to form a function unit with an independent variable of repetition frequency j and a function value of reference power;
all the function units form the binary piecewise function prediction model.
9. A self-learning laser power fluctuation identification system as claimed in claim 3, wherein: the forming of the function unit is specifically:
setting the functional relation between the reference power and the repetition frequency j as follows:
P＝f(j)＝a*lnj+b
define the sum of the squares of the total errors:
Figure FDA0003107537270000031
adding Se2Respectively solving partial derivatives of a and b;
adding Se2Taking the minimum value and associating each repetition frequency j with its corresponding average power P (A)ij) And (4) solving a and b and putting the values into the functional relation.
10. A self-learning laser power fluctuation identification system according to any one of claims 6 to 9, wherein: the device also comprises a power alarm module (9); the power alarm module (9) comprises a comparison unit (10) and an alarm unit (11);
the comparison unit (10) is configured to acquire the actual power and the reference power of the laser to be detected and calculate a fluctuation difference value; and the alarm unit (11) is configured to send alarm information when the fluctuation difference value is judged to be larger than a set value.
CN202110642357.XA 2021-06-09 2021-06-09 Self-learning laser power fluctuation identification method and system Active CN113378462B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110642357.XA CN113378462B (en) 2021-06-09 2021-06-09 Self-learning laser power fluctuation identification method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110642357.XA CN113378462B (en) 2021-06-09 2021-06-09 Self-learning laser power fluctuation identification method and system

Publications (2)

Publication Number Publication Date
CN113378462A true CN113378462A (en) 2021-09-10
CN113378462B CN113378462B (en) 2022-06-17

Family

ID=77573214

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110642357.XA Active CN113378462B (en) 2021-06-09 2021-06-09 Self-learning laser power fluctuation identification method and system

Country Status (1)

Country Link
CN (1) CN113378462B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050018723A1 (en) * 2003-05-14 2005-01-27 Masayuki Morita Method of stabilizing laser beam, and laser beam generation system
CN108844624A (en) * 2018-06-01 2018-11-20 北京科技大学 A kind of SLM process laser power monitor method based on temperature field
CN109687479A (en) * 2017-10-19 2019-04-26 中国南方电网有限责任公司 Power swing stabilizes method, system, storage medium and computer equipment
CN110070226A (en) * 2019-04-24 2019-07-30 河海大学 Photovoltaic power prediction technique and system based on convolutional neural networks and meta learning
CN110429466A (en) * 2019-06-24 2019-11-08 东莞理工学院 A kind of high-power semiconductor laser real-time detecting system
CN110579336A (en) * 2019-10-15 2019-12-17 核工业理化工程研究院 Laser amplification chain output power prediction system and method
CN112326197A (en) * 2020-10-23 2021-02-05 中国科学院上海光学精密机械研究所 Method for predicting long service life of laser optical component

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050018723A1 (en) * 2003-05-14 2005-01-27 Masayuki Morita Method of stabilizing laser beam, and laser beam generation system
CN109687479A (en) * 2017-10-19 2019-04-26 中国南方电网有限责任公司 Power swing stabilizes method, system, storage medium and computer equipment
CN108844624A (en) * 2018-06-01 2018-11-20 北京科技大学 A kind of SLM process laser power monitor method based on temperature field
CN110070226A (en) * 2019-04-24 2019-07-30 河海大学 Photovoltaic power prediction technique and system based on convolutional neural networks and meta learning
CN110429466A (en) * 2019-06-24 2019-11-08 东莞理工学院 A kind of high-power semiconductor laser real-time detecting system
CN110579336A (en) * 2019-10-15 2019-12-17 核工业理化工程研究院 Laser amplification chain output power prediction system and method
CN112326197A (en) * 2020-10-23 2021-02-05 中国科学院上海光学精密机械研究所 Method for predicting long service life of laser optical component

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
Z.L. LU ET AL.: "The prediction of the building precision in the Laser Engineered Net shaping process using advanced networks", 《OPTICS AND LASER IN ENGINEERING》 *

Also Published As

Publication number Publication date
CN113378462B (en) 2022-06-17

Similar Documents

Publication Publication Date Title
US10078495B2 (en) Method and apparatus for generating source-independent quantum random number
CN101995225B (en) Film thickness measurement apparatus
CN103913211B (en) Time coefficient calibration method of ultrasonic water meter
CN111476430A (en) Tool residual life prediction method based on machine learning regression algorithm
US20140129503A1 (en) Method for predicting machining quality of machine tool
CN105181644A (en) Online monitoring system and method for cigarette paper
CN111898443B (en) Flow monitoring method for wire feeding mechanism of FDM type 3D printer
CN113378462B (en) Self-learning laser power fluctuation identification method and system
CN107064753A (en) Bow net arc-plasma Multi-parameter Data Acquisition method and apparatus
KR20120000228A (en) Partial discharging detector and method for detecting the partial discharging
CN106645952A (en) Signal phase difference detection method and system
CN107688820B (en) Elevator fault diagnosis method based on BCSA optimized support vector machine
CN109934334B (en) Disturbance-based chlorophyll a content related factor sensitivity analysis method
CN117235547B (en) Self-adaptive filtering method for oxygen concentration detection data
CN103559417A (en) Intelligent soft measurement method of slashing sizing percentage
CN107808209B (en) Wind power plant abnormal data identification method based on weighted kNN distance
CN103093078A (en) Data inspection method for improved 53H algorithm
CN102726833A (en) Method for determining reliability of cigarette maker cigarette loose end detector
CN112801328B (en) Product printing parameter setting device, method and computer readable storage medium
CN105300868B (en) A kind of hole-punching huon pine paper air permeability detection method in tobacco business
KR101945131B1 (en) Method and Apparatus for Managing Very Small Fraction of Nonconforming under Non-Normal Process
IL309308A (en) Signal-to-noise-ratio metric for determining nucleotide-base calls and base-call quality
US20130282332A1 (en) Method of obtaining linear curve fitting conversion equation for use with non-linear measurement system
CN105097589B (en) A kind of detection method of metal hardmask integration etching through hole over etching amount
US20180025894A1 (en) Plasma processing apparatus and analysis method for analyzing plasma processing data

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant