Summary of the invention
Goal of the invention: aiming at the problems existing in the prior art, the present invention, which provides one kind, can quickly and effectively identify event
Hinder welding product, improve production efficiency and ensures the anchor chain flash welding quality online evaluation method of welding quality.
Technical solution: in order to solve the above technical problems, the present invention provides a kind of anchor chain flash welding quality online evaluation method,
It is characterized by comprising the following steps:
(1) L normal weld 2D signal composition data library is chosen;
(2) data in step (1) are normalized;
(3) signal subsection is handled;
(4) the dissimilar distance between L signal is calculated;
(5) the dissimilar distance average g between L signal is calculated;
(6) judge whether system monitors new measured signal, be judged as YES, then follow the steps (7);It is judged as NO, then
Indicate that welding stops, system finishing process;
(7) measured signal is acquired;
(8) selection standard signal;
(9) data in step (7) and step (8) are normalized and signal subsection processing;
(10) dissmilarity distance g between measured signal and standard signal is calculated;
(11) judge whether dissmilarity distance g is less than according to step (5) and step (10)Wherein α is set by user
Quality evaluation sensitivity coefficient, if the judgment is Yes, g is less thanThen measured signal is normal welding signal, is executed step (12),
If the judgment is No, g is more than or equal toThen measured signal is abnormal solder signal, and system issues early warning and terminates to flow
Journey;
(12) judge that measured signal for normal welding signal, is handled normal signal according to step (11).
Further, in the step (2) data in step (1) are normalized with specific step is as follows:
The L normal weld signal wherein chosen in step (1) is E1,E2,,EL, wherein L selected signal be by
The 2D signal of electrode position signal and current signal composition, such as E1=[A B], A=[a1,a2,,ai,,an]TFor electrode position
Confidence number, B=[b1,b2,,bi,bn]TFor current signal;
According to z-score algorithm to electrode position and electric current 2D signal E1It is normalized:
Wherein μ1For the mean value of sample data A, σ1For the standard deviation of sample data A, A*For electrode position signal A normalization
Result that treated.μ2For the mean value of sample data B, σ2For the standard deviation of sample data B, B*For current signal B normalized
Result afterwards;2D signal after obtaining normalizedL signal is carried out respectively according to above-mentioned formula
Normalized, the signal after being normalized
Further, signal subsection is handled in the step (3), according to welding stage butt welding different during actual welding
Connecing quality influences difference, to the 2D signal after normalizationSegment processing is carried out, warm-up phase X1, group flashing light rank are divided into
Section X2 and upset stage X3;To the 2D signal after normalizationSegment processing is carried out, warm-up phase Y1, group flashing light are divided into
Stage Y2 and upset stage Y3 is similarly rightCarry out segment processing.
Further, specific step is as follows for the dissimilar distance in the step (4) between L signal of calculating:
(4.1) it calculatesWithBetween dissimilar distance g1,2, g1,2=ρ1g1+ρ2g2+ρ3g3, wherein ρ1,ρ2,ρ3To be
Number, g1,g2,g3RespectivelyWithIn warm-up phase, the dissimilar distance in group flashing light stage, upset stage;
(4.2) L 2D signal is calculated separately according to above-mentioned steps 4.1Between dissimilar distance gi,j
(i=1,2, L, j=2,3, L).
Further, it is calculated in the step (4.1)WithBetween dissimilar distance g1,2, g1,2=ρ1g1+ρ2g2+
ρ3g3Specific step is as follows:
(4.1.1) is for 2D signalWithIn the data X1 and Y1 of warm-up phase, first construction Distance matrix D=
[di,j], wherein element di,jIndicate the Euclidean distance between X1 (i) and Y1 (i):
di,j=| X1 (i)-Y1 (j) |, i=1,2, n, j=1,2, m
N and m is the length of X1 and Y1 respectively;
(4.1.2) searches for the path W={ w of a connection (1,1) and (n, m) in a two-dimensional matrix1,w2,,wk,
In, k is the total length in path, and the value of k is determined by solving optimal path W, i.e. w1=(1,1) and wk=(n, m), together
When meet monotonicity and step-length and be less than the two constraint conditions of r;
(4.1.3) finds optimal regular path, from primary condition θ (1, j)=d1,j=| X1 (1)-Y1 (j) |, j=1,2,
M, θ (i, 1)=di,1=| X1 (i)-Y1 (1) |, i=1,2, n starts, and step-length is less than r, and searching algorithm is as follows:
Wherein θ (i-1, j-1), θ (i-1, j) and θ (i, j-1) indicate three lattice points (i-1, j-1) that may advance, (i-
1, j) it is indicated in current Cumulative Distance with the Cumulative Distance of (i, j-1), min (θ (i-1, j-1), θ (i-1, j), θ (i, j-1))
Minimum value;θ (i, j) is minimum Cumulative Distance and current lattice point distance di,jThe sum of, total Cumulative Distance as current lattice point;
(4.1.4) finally calculates dissimilar distance are as follows:
g1Indicate two 2D signalsWithIn the dissimilar distance of the data X1 and Y1 of warm-up phase;
(4.1.5) calculates separately 2D signal according to step (4.1.1)-(4.1.4)WithIn the group flashing light stage and
The dissimilar distance g in upset stage1, g2。
Further, the dissimilar distance average between L signal is calculated in the step (5)Specific steps such as
Under:
Further, specific step is as follows for acquisition measured signal in the step (7):
The electrode position signal Q=[q generated in acquisition welding process1,q2,,qi,,qn]T, current signal C=[c1,c2,,
ci,,cn]T, sampling time T1, sampling interval t1, n=T1/t1;Electrode position signal Q and current signal C is formed one
Two-dimentional measured signal S=[Q C].
Further, specific step is as follows for first selection standard signal in the step (8):
According to actual welding experience selection standard signal F=[V Z], V=[v1,v2,,vi,,vm]TFor electrode position signal,
Z=[z1,z2,,zi,,zm]TFor current signal, sampling time T2, sampling interval t2, m=T2/t2;According to step (2) to mark
Calibration signal F is normalized, the signal F after obtaining normalized*=[V* Z*]。
Further, handled normal signal that specific step is as follows in the step (12):
It rejects the signal being added at first in L data field signal and (is generally defaulted as E1) and this measured signal S is added, it returns
Step (1) is gone back to be updated database.
Compared with the prior art, the advantages of the present invention are as follows:
A kind of anchor chain flash welding quality online evaluation method of the invention, with simple algorithm by the identification to failure weldment
It is converted to the signal for acquiring sensor in welding process to analyze, eliminates complicated modeling process and a large amount of meter
It calculates, the space-time dissimilarity between measured signal and standard signal can be quantified, realize pair quick and compared with high-accuracy
Anchor chain Flash Butt Welding quality carries out online evaluation.It is proposed by the present invention that welding signal is subjected to segment processing, adequately consider
The influence of different welding stage welding qualities.And it changes with time to weaken welding system performance and welding machine state
To the influence that the welding signal of acquisition generates, the real-time update to database is realized.It can be with the identification compared with high-accuracy simultaneously
And the abnormality of real-time early warning welding, find the potential quality problems in anchor chain welding process, in time for the height for ensuring anchor chain
Quality and navigation safety provide reliable guarantee.
Embodiment 1:
Step 1) welds in record data from history according to actual welding experience and chooses 100 normal weld signal E1,E2,,
E100(2D signal that 100 selected signals are made of electrode position signal and current signal, such as E1=[A B], A
=[a1,a2,,ai,,an]TFor electrode position signal, B=[b1,b2,,bi,bn]TFor current signal).
Step 2) data prediction.With z-score algorithm to electrode position and electric current 2D signal E1,E2,,E100Returned
One change processing, the signal after being normalized
The processing of step 3) signal subsection.Difference is influenced according to welding stage welding qualities different during actual welding,
To the 2D signal after normalizationSegment processing is carried out, warm-up phase, group flashing light stage and upset are divided into
Stage.
Step 4) dtw algorithm calculates the 2D signal after 100 segment processingsNot phase between any two
Like distance.For example,WithBetween dissimilar distance g1,2, g1,2=ρ1g1+ρ2g2+ρ3g3, according to different welding stage butt welding
Meet the different setting coefficient ρ that quality has an impact1=0.35, ρ2=0.5, ρ3=0.15, g1,g2,g3RespectivelyWithPre-
The dissimilar distance in hot stage, group flashing light stage, upset stage.
Step 5) calculates the dissimilar distance average between 100 signals
Step 6) judges whether system monitors new measured signal, is judged as YES, and thens follow the steps (7);It is judged as NO,
Then indicate that welding stops, system finishing process;
The electrode position signal Q generated in step 7) acquisition welding process1With current signal C1, form a 2D signal
S1(as shown in Fig. 2), sampling time 70.8s, sampling interval 0.1s.
Step 8) is made of according to actual welding experience selection standard signal F, F electrode position signal V and current signal Z
2D signal (as shown in Fig. 2), sampling time 69.1s, sampling interval 0.1s.
Step 9) S to measured signal1It is normalized to obtain S1 *=[Q1 * C1 *] (as shown in Fig. 2).To standard
Signal F is normalized to obtain F*=[V* Z*] (as shown in Fig. 2).According to the welding stages different during actual welding
Welding quality influences difference, to the measured signal S after normalization1 *Segment processing is carried out, warm-up phase P1, group flashing light are divided into
Stage P2 and upset stage P3.To the standard signal F after normalization*Segment processing is carried out, warm-up phase H1, group flashing light are divided into
Stage H2 and upset stage H3 (as shown in Fig. 2).
Step 10) chooses step-length r=70, calculates the dissmilarity between the measured signal P1 of warm-up phase and standard signal H1
Distance g1=23.35, calculate the dissimilar distance g between the measured signal P2 and standard signal H2 in group flashing light stage2=
21.02, calculate the dissimilar distance g between the measured signal P3 and standard signal H3 in upset stage3=4.56.Coefficient ρ is set1
=0.35, ρ2=0.5, ρ3=0.15, the dissmilarity distance g=ρ between measured signal and standard signal1g1+ρ2g2+ρ3g3=
19.37。
Step 11) sets quality evaluation sensitivity coefficient α=1, judges whether dissmilarity distance g is less thanAccording to step 5
With step 10 as a result, judgement(as shown in Fig. 3), then measured signal is normal welding signal, executes step 12.
Step 12) judges that measured signal for normal welding signal, is handled normal signal according to step 11.It rejects
The signal being added at first in 100 data field signals (is generally defaulted as E1) and this measured signal S is added1, return step 1 is right
Database is updated.
Embodiment 2:
Step 1) welds in record data from history according to actual welding experience and chooses 100 normal weld signal E1,E2,,
E100(2D signal that 100 selected signals are made of electrode position signal and current signal, such as E1=[A B], A
=[a1,a2,,ai,,an]TFor electrode position signal, B=[b1,b2,,bi,bn]TFor current signal).
Step 2) data prediction.With z-score algorithm to electrode position and electric current 2D signal E1,E2,,E100Returned
One change processing, the signal after being normalized
The processing of step 3) signal subsection.Difference is influenced according to welding stage welding qualities different during actual welding,
To the 2D signal after normalizationSegment processing is carried out, warm-up phase, group flashing light stage and upset rank are divided into
Section.
Step 4) dtw algorithm calculates the 2D signal after 100 segment processingsNot phase between any two
Like distance.For example,WithBetween dissimilar distance g1,2, g1,2=ρ1g1+ρ2g2+ρ3g3, according to different welding stage butt welding
The difference that quality has an impact is connect, coefficient ρ is set1=0.35, ρ2=0.5, ρ3=0.15, g1,g2,g3RespectivelyWith?
Warm-up phase, the dissimilar distance in group flashing light stage, upset stage.
Step 5) calculates the dissimilar distance average between 100 signals
Step 6) judges whether system monitors new measured signal, is judged as YES, and thens follow the steps (7);It is judged as NO,
Then indicate that welding stops, system finishing process;
The electrode position signal Q generated in step 7) acquisition welding process2With current signal C2, form a 2D signal
S2(as shown in Fig. 2), sampling time 89.5s, sampling interval 0.1s.
Step 8) is made of according to actual welding experience selection standard signal F, F electrode position signal V and current signal Z
2D signal (as shown in Fig. 2), sampling time 69.1s, sampling interval 0.1s.
Step 9) S to measured signal2It is normalized to obtain S2 *=[Q2 * C2 *] (as shown in Fig. 2).To standard
Signal F is normalized to obtain F*=[V* Z*] (as shown in Fig. 2).According to the welding stages different during actual welding
Welding quality influences difference, to the measured signal S after normalization2 *Segment processing is carried out, warm-up phase P1, group flashing light are divided into
Stage P2 and upset stage P3.To the standard signal F after normalization*Segment processing is carried out, warm-up phase H1, group flashing light are divided into
Stage H2 and upset stage H3 (as shown in Fig. 2).
Step 10) chooses step-length r=70, calculates the dissmilarity between the measured signal P1 of warm-up phase and standard signal H1
Distance g1=46.13, calculate the dissimilar distance g between the measured signal P2 and standard signal H2 in group flashing light stage2=
39.23, calculate the dissimilar distance g between the measured signal P3 and standard signal H3 in upset stage3=6.94.Coefficient ρ is set1
=0.35, ρ2=0.5, ρ3=0.15, the dissmilarity distance g=ρ between measured signal and standard signal1g1+ρ2g2+ρ3g3=
36.8。
Step 11) setting quality evaluation sensitivity coefficient α=1 judges whether dissmilarity distance g is less thanAccording to step 5 and
Step 10 as a result, judgement(as shown in Fig. 3), then measured signal is abnormal solder signal, and system issues early warning simultaneously
Terminate process.
The principles and effects of the invention, and the implementation that part uses only is illustrated in the above embodiments
Example, and is not intended to limit the present invention;It should be pointed out that for those of ordinary skill in the art, not departing from wound of the present invention
Under the premise of making design, various modifications and improvements can be made, and these are all within the scope of protection of the present invention.