CN103394795B - Method for adaptively detecting periodic phases of double-pulse welding current waveforms - Google Patents

Abstract

The invention discloses a self-adaptive detection method for the cycle phase of the double-pulse welding current waveform, which comprises the following steps: collecting a current signal I of a double-pulse welding welding process; and estimating the local Time range; calculate the current mean value of each point in the current signal I within the local time range to form a current local mean value sequence; determine the strong pulse group and weak pulse group stage of each point in the current local mean value sequence, and obtain the group Phase mark sequence; correct each group jump edge in the group stage mark; based on the group stage mark sequence, within the range of each group, mark the period phases of all signal points in the group. The self-adaptive detection method of the cycle phase of the double pulse welding current waveform of the present invention can detect the mark of the cycle phase more accurately for the double pulse welding of various modulation modes, and lays the foundation for subsequent signal analysis and processing.

Description

An Adaptive Detection Method for Periodic Stages of Double-Pulse Welding Current Waveform

technical field

The invention relates to an automatic detection method for arc welding, in particular to an adaptive detection method for the cycle phase of a double-pulse welding current waveform, which belongs to the field of arc welding detection.

Background technique

With the wide application of arc welding, the performance of arc welding power source has become one of the key points affecting the quality of industrial products. Throughout _CO2 welding, single pulse welding, double pulse welding and other welding methods, the performance of arc welding power supply is very important, which affects the quality, efficiency and cost of welding. Therefore, it is particularly important to evaluate the performance of arc welding power supply. This is mainly realized by detecting the voltage and current signals output by the arc welding power source. To carry out in-depth detection and analysis, the prerequisite step is to mark the cycle phase of each signal point of the voltage or current waveform.

Double-pulse welding is to use low-frequency pulses to modulate the peak current and peak time of the high-frequency droplet transfer pulse, so that the intensity of the unit pulse is periodically switched between strong and weak at low frequency, and periodic changes in strong and weak pulses are obtained. group; while achieving a beautiful fish-scale weld appearance, it ensures high welding efficiency, reduces the incidence of pores, and refines the weld grains.

Double pulse welding controls the welding current, and the cycle phase information is implicit in the current waveform. Each sampling point of the welding current may be in one of the four cycle phases of strong pulse group peak value stage, strong pulse group base value stage, weak pulse group peak value stage and weak pulse group base value stage. Only by accurately distinguishing the period phase, it is possible to further calculate a series of characteristic parameters such as the average value of the peak value of the strong burst and the time average value of the peak value of the strong burst.

However, due to the multiple modulation methods of double pulse welding, such as changing the peak current size, changing the base value current size, changing the duty cycle of high-frequency pulses, and realizing slow-change conversion between strong pulse groups and weak pulse groups, etc., Even if it is a fixed form, its parameters are adjustable, so it is difficult to divide the cycle phases of the double-pulse welding current waveform, which in turn has a greater impact on the subsequent signal analysis and processing. For example, the inaccurate division of the period phase will cause large errors in a series of characteristic parameters such as the average value of the peak value of the strong burst and the time average value of the peak value of the strong burst.

Contents of the invention

The purpose of the present invention is to solve the defects of the above-mentioned prior art, and to provide an adaptive detection method for the period stage of the double-pulse welding current waveform. Periodic phase marks, laying the foundation for subsequent signal analysis and processing.

The purpose of the present invention can be achieved by taking the following technical solutions:

The self-adaptive detection method of the period stage of the double pulse welding current waveform adopts a test platform based on an industrial computer and an arc welding process detector. The arc welding process detector includes a Hall current sensor, and is characterized in that it includes the following steps:

1) Utilize the Hall current sensor to detect the welding current of a double-pulse welding process, and sample through the data acquisition card, and output the obtained current signal I to the industrial computer;

2) Estimate the local time range according to the average value and standard deviation of the current signal I;

3) Calculate the current mean value of each point in the current signal I within the local time range, thereby forming a current local mean value sequence;

4) The double-threshold method is used to determine the strong burst and weak burst stages of each point of the current local mean sequence, and the group stage marker sequence is obtained;

5) Based on the group stage marking sequence, within the scope of each group, use the double threshold method to mark the period stage of all signal points in the group.

As a preferred solution, the estimated local time range described in step 2) is as follows:

2.1) Set a threshold swing parameter, and calculate the upper threshold, middle threshold and lower threshold of the current through the following formulas (1), formula (2) and formula (3):

uV＝avg+std×q (1)

mV＝avg (2)

dV＝avg-std×q (3)

Wherein, avg is the average value of the current signal I, std is the standard deviation of the current signal I, q is the threshold swing parameter, uV is the upper threshold of the current, mV is the middle threshold of the current, and dV is the lower threshold of the current;

2.2) Use the three threshold values of uV, mV and dV to count the average values uP, mP and dP of the current signal I from 1 to 1000 cycles;

2.3) Take the median of uP, mP and dP as the reference period;

2.4) Take 0.5-1000 times of the reference period as the local time range.

As a preferred solution, the range of the threshold swing parameter q is (0, 100).

As a preferred solution, step 4) adopts the double threshold method to determine the strong burst and weak burst stages of each point of the current local mean sequence, as follows:

4.1) Calculate the mean and standard deviation of the current local mean sequence; set a threshold swing parameter of the current local mean sequence, calculate the upper threshold and Lower threshold:

ULV＝AVG+STD×SR (4)

DLV＝AVG-STD×SR (5)

Among them, AVG is the average value of the current local average sequence, STD is the standard deviation of the current local average sequence; SR is the threshold swing parameter of the current local average sequence, ULV is the upper threshold of the current local average sequence, and DLV is the current The lower threshold of the local mean sequence;

4.2) If the value of the first point of the current local mean sequence is greater than or equal to ULV, mark this point as a strong burst phase; if the value of the first point is less than or equal to DLV, mark this point as a weak burst phase; otherwise , that is, the value of the first point is greater than DLV and less than ULV, marking this point as a weak burst stage;

4.3) If the value of the i-th point of the current local mean sequence is greater than or equal to ULV, mark the i-th point as a strong burst phase; if the value of the i-th point is less than or equal to DLV, mark the i-th point as a weak burst phase ; Otherwise, that is, the value of the i-th point is greater than DLV and less than ULV, the i-th point inherits the stage mark of the i-1th point; where, i≥2.

As a preferred solution, the range of the threshold swing parameter SR of the current local average value sequence is (0, 100).

As a preferred solution, before step 5), each strong and weak pulse group jump edge point in the correction group stage marker sequence is also included, specifically as follows:

For any edge point k of the hth strong and weak pulse group jump in the group stage marker sequence, the value of the kth point of the current local average value sequence is taken as the threshold value, and the current signal I takes the kth point as the center forward and toward After detecting the high-frequency pulse jump, find the nearest high-frequency pulse jump point u, replace k, and correct it as the h-th strong and weak pulse group jump edge point.

As a preferred solution, step 5) adopts the double threshold method to mark the periodic stage, as follows:

5.1) For any strong pulse group or weak pulse group starting from the mth point and ending at the nth point in the group stage marker sequence, calculate the mean value and standard deviation of the current signal I within the range of [m, n], as follows Formulas (6) and (7) calculate the upper and lower thresholds of the current in the group:

groupULV＝groupAvg+groupStd×groupSR (6)

groupDLV＝groupAvg-groupStd×groupSR (7)

Among them, groupAvg is the average value of the current signal I within the range of [m, n], groupStd is the standard deviation of the current signal I within the range of [m, n], groupSR is the swing parameter of the current threshold within the group, and groupULV is the upper limit of the current within the group Threshold, groupDLV is the lower threshold of current in the group;

5.2) If the value of the mth point of the current signal I is greater than or equal to groupULV, then mark this point as the peak stage in the group; if the value of the mth point is less than or equal to groupDLV, then mark this point as the base value stage in the group; Otherwise, that is, the value of the mth point is greater than groupDLV and less than groupULV, mark this point as the base value stage in the group;

5.3) If the mth point of the current signal I is in a strong pulse group and is the peak stage within the group, then mark the period phase of this point as "strong burst peak value"; In the base value stage, mark the periodic stage of this point as "strong burst base value"; if it is a weak burst and it is the peak stage in the group, then mark the periodic stage of this point as "weak burst peak value"; if If the pulse group is weak and is the base value stage within the group, then mark the periodic stage of this point as "weak burst base value";

5.4) If the value of the m+p point of the current signal I is greater than or equal to groupULV, mark the m+p point as the peak stage in the group; if the value of the m+p point is less than or equal to groupDLV, mark the m+ Point p is the base value stage in the group; otherwise, that is, the value of the m+p point is greater than groupDLV and less than groupULV, and the m+p point inherits the group mark of the m+p-1 point; among them, 1≤p≤ n-m;

5.5) If the m+p point of the current signal I is in a strong pulse group, and it is the peak stage in the group, then mark the period stage of this point as "strong burst peak"; if it is a strong pulse group and is a group If it is the base value stage within the group, mark the periodic stage of this point as "strong burst base value"; if it is a weak burst and it is the peak stage within the group, then mark the periodic stage of this point as "weak burst peak value" ; If the pulse group is weak and it is the base value stage in the group, then mark the periodic stage of this point as "weak burst group base value".

As a preferred solution, the range of the current threshold swing parameter groupSR within the group is (0, 100).

As a preferred solution, the test platform also includes a welding wire conveying mechanism, a trolley, a guide rail and an oscilloscope; the arc welding process detector also includes a voltage sensor.

Compared with the prior art, the present invention has the following beneficial effects:

1. The self-adaptive detection method of the double-pulse welding current waveform cycle stage of the present invention uses the low-frequency modulation of double-pulse welding to change the magnitude of the local current periodically to realize the periodicity of the instantaneous energy obtained by the welding wire and the molten pool The characteristics of the control can detect the double pulse welding of various modulation modes. This method can accurately distinguish the strong pulse group and the weak pulse group stage by detecting the change of the current average value in the local time range, and further divide them into groups within the peak and base phases.

2. In the self-adaptive detection method of the cycle stage of the double-pulse welding current waveform of the present invention, the calculated local mean value of the current is larger when the pulse group is strong, and smaller when the pulse group is weak, and has a good correlation with the pulse group stage ; The group phase marks obtained by analyzing the double-threshold values of the current local average sequence are very consistent with the changes of the original current waveform; moreover, most of the high-frequency pulse cycle marks in the pulse group are also correct. This makes the error less when further calculating a series of characteristic parameters such as the average value of the peak value of the strong burst and the time average value of the peak value of the strong burst, which can be said to provide convenience for subsequent signal analysis and processing.

Description of drawings

Fig. 1 is a schematic flow chart of the self-adaptive detection method for the periodic phase of the double-pulse welding current waveform of the present invention.

Figures 2a-2c are schematic diagrams of the first double-pulse welding current waveform, its local mean sequence waveform and pulse group stage marks, respectively.

Figures 3a-3b are schematic diagrams of the first type of double-pulse welding current waveform and cycle stage marks after enlarged display respectively.

Figures 4a to 4c are schematic diagrams of the second double-pulse welding current waveform, its local mean value sequence waveform, and pulse group stage marks, respectively.

Figures 5a to 5c are schematic diagrams of the third double-pulse welding current waveform, its local mean value sequence waveform, and pulse group stage marks, respectively.

Figures 6a to 6c are schematic diagrams of the fourth double-pulse welding current waveform, its local mean value sequence waveform, and pulse group stage marks, respectively.

Figures 7a to 7c are schematic diagrams of the fifth double-pulse welding current waveform, its local average value sequence waveform, and pulse group stage marks, respectively.

Figures 8a to 8c are schematic diagrams of the sixth double-pulse welding current waveform, its local mean value sequence waveform, and pulse group stage marks, respectively.

Figures 9a-9b are schematic diagrams of the second double-pulse welding current waveform and cycle stage marks after enlarged display respectively.

Figures 10a-10b are schematic diagrams of the third double-pulse welding current waveform and cycle stage marks after enlarged display respectively.

Figures 11a-12b are schematic diagrams of the fourth double-pulse welding current waveform and cycle stage marks after enlarged display respectively.

Figures 12a-12b are schematic diagrams of the fifth double-pulse welding current waveform and cycle stage marks after enlarged display respectively.

Figures 13a-13b are schematic diagrams of the sixth double-pulse welding current waveform and cycle stage marks after enlarged display respectively.

Detailed ways

Example 1:

The self-adaptive detection method of the double-pulse welding current waveform period stage of the present embodiment adopts a test platform including an industrial computer, an arc welding process detector, a welding wire conveying mechanism, a walking trolley, a guide rail and an oscilloscope, and the arc welding process detector includes a voltage sensor and Hall current sensor.

As shown in Figure 1, the adaptive detection method of the double pulse welding current waveform period stage of the present embodiment includes the following steps:

1) Use the Hall current sensor to detect the welding current of a double-pulse welding process, and sample it through the data acquisition card, and output the obtained current signal I to the industrial computer. The waveform displayed by the oscilloscope is shown in Figure 2a. The current waveform The characteristics are: the current peak value of the strong and weak pulse group is different, and the base value is also different;

2) Calculate the average value and standard deviation of the current in the double-pulse arc welding process according to the current signal I, and then estimate the local time range. The local time range is a sliding time window commonly used in signal processing, as follows:

2.1) Set a threshold swing parameter, and calculate the upper threshold, middle threshold and lower threshold of the current through the following formulas (1), formula (2) and formula (3):

uV＝avg+std×q (1)

mV＝avg (2)

dV＝avg-std×q (3)

Among them, avg is the average value of the current in the double-pulse arc welding process, std is the standard deviation of the current in the double-pulse arc welding process, q is the threshold swing parameter with a value of 0.5, uV is the upper threshold of the current, mV is the middle threshold of the current, dV is the lower threshold of the current;

2.2) Calculate the average value uP, mP and dP of 50 cycles of the current signal I with three threshold values of uV, mV and dV respectively;

2.3) Take the median of uP, mP and dP as the reference period;

2.4) Take twice the reference period as the local time range;

3) Calculate the current mean value of each point in the current signal I within the local time range, thereby forming a current local mean value sequence, as shown in Figure 2b;

4) The double-threshold method is used to determine the strong burst and weak burst phases of each point of the current local mean sequence, and the group stage marker sequence is obtained, as follows:

4.1) Calculate the mean and standard deviation of the current local mean sequence; set a threshold swing parameter of the current local mean sequence, calculate the upper threshold and Lower threshold:

ULV＝AVG+STD×SR (4)

DLV＝AVG-STD×SR (5)

Among them, AVG is the average value of the current local average sequence, STD is the standard deviation of the current local average sequence; SR is the threshold swing parameter of the current local average sequence with a value of 0.5, and ULV is the upper threshold of the current local average sequence , DLV is the lower threshold of the current local mean sequence;

4.2) If the value of the first point of the current local mean sequence is greater than or equal to ULV, mark this point as a strong burst phase; if the value of the first point is less than or equal to DLV, mark this point as a weak burst phase; otherwise , that is, the value of the first point is greater than DLV and less than ULV, marking this point as a weak burst stage;

4.3) If the value of the i-th point of the current local mean sequence is greater than or equal to ULV, mark the i-th point as a strong burst phase; if the value of the i-th point is less than or equal to DLV, mark the i-th point as a weak burst stage; otherwise, that is, the value of the i-th point is greater than DLV and less than ULV, the i-th point inherits the stage mark of the i-1th point; where, i≥2;

5) For any edge point k of the h-th strong and weak pulse group jump in the group stage marker sequence, the value of the k-th point in the current local mean sequence is the threshold value, and the current signal I is centered on the k-th point forward And detect the jump of high-frequency pulse backwards, find the nearest high-frequency pulse jump point u, replace k, and correct it as the edge point of the h-th strong and weak pulse group jump; the finally obtained pulse group stage mark is shown in the figure as shown in 2c;

6) Based on the group stage marking sequence, within the scope of each group, use the double threshold method to mark the periodic stage of all signal points in the group, as follows:

6.1) For any strong pulse group or weak pulse group starting from the mth point and ending at the nth point in the group stage marker sequence, calculate the mean value and standard deviation of the current signal I within the range of [m, n], as follows Formulas (6) and (7) calculate the upper and lower thresholds of the current in the group:

groupULV＝groupAvg+groupStd×groupSR (6)

groupDLV＝groupAvg-groupStd×groupSR (7)

Among them, groupAvg is the average value of the current signal I within the range of [m, n], groupStd is the standard deviation of the current signal I within the range of [m, n], groupSR is the current threshold swing parameter in the group with a value of 0.5, and groupULV is the group The upper threshold of the internal current, groupDLV is the lower threshold of the current within the group;

6.2) If the value of the mth point of the current signal I is greater than or equal to groupULV, then mark this point as the peak stage in the group; if the value of the mth point is less than or equal to groupDLV, then mark this point as the base value stage in the group; Otherwise, that is, the value of the mth point is greater than groupDLV and less than groupULV, mark this point as the base value stage in the group;

6.3) If the mth point of the current signal I is in a strong pulse group and is the peak stage within the group, then mark the period phase of this point as "strong burst peak"; In the base value stage, mark the periodic stage of this point as "strong burst base value"; if it is a weak burst and it is the peak stage in the group, then mark the periodic stage of this point as "weak burst peak value"; if If the pulse group is weak and is the base value stage within the group, then mark the periodic stage of this point as "weak burst base value";

6.4) If the value of the m+p point of the current signal I is greater than or equal to groupULV, mark the m+p point as the peak stage in the group; if the value of the m+p point is less than or equal to groupDLV, then mark the m+p point Point p is the base value stage in the group; otherwise, that is, the value of the m+p point is greater than groupDLV and less than groupULV, and the m+p point inherits the group mark of the m+p-1 point; among them, 1≤p≤ n-m;

6.5) If the m+p point of the current signal I is in a strong pulse group, and it is the peak stage in the group, then mark the period phase of this point as "strong burst peak"; if it is a strong pulse group and is a group If it is the base value stage within the group, mark the periodic stage of this point as "strong burst base value"; if it is a weak burst and it is the peak stage within the group, then mark the periodic stage of this point as "weak burst peak value" ; If the pulse group is weak and it is the base value stage in the group, then mark the periodic stage of this point as "weak burst group base value".

As shown in Figure 3a~3b, it is the enlarged display of the current waveform and the cycle stage marking diagram, "2" represents the peak stage of the strong burst, "1" represents the base value stage of the strong burst, and "-1" represents the weak burst In the peak stage, "-2" represents the base value stage of the weak pulse group; most of the marks in the cycle stage are correct, and the peak or base value stage marks in the group are consistent with the original current waveform, and there is a certain rate of error at the edge of the group jump. However, the error brought about by the statistical calculation of subsequent eigenvalues is very small, because in general, a low-frequency pulse group contains more high-frequency pulse periods.

Example 2:

The main features of this embodiment are: the self-adaptive detection method of the above-mentioned embodiment 1 is used to detect the periodic stages of the other five double-pulse welding process currents. The various characteristics of the five double-pulse welding current waveforms are: the second one is shown in Figure 4a As shown, the current peak value of the strong and weak pulse group is the same, but the base value is different; the third type is shown in Figure 5a, the current peak value of the strong and weak pulse group is different, and the base value is the same; the fourth type is shown in Figure 6a, the strong and weak pulse group The current peak value of the group is similar, and the base value is also similar, but the duty ratio of the high-frequency pulse of the strong and weak pulse group is different; As shown in Fig. 8a, the edges of strong and weak bursts transition smoothly.

For each of the above double-pulse welding, the current signal of an arc welding process is collected respectively, and the current local mean value sequence is calculated, as shown in Figure 4b-8b, the current local mean value is larger when the strong pulse group is strong, and the current local mean value is larger when the weak pulse group is weak. It is small and has a good correlation with the pulse group stage; as shown in Figures 4c to 8c, the group stage markers obtained through the double threshold analysis of the current local mean sequence waveform are very consistent with the changes in the current waveforms in Figures 4a to 8a.

Finally, the base value or peak stage in the marker group, as shown in Figures 9a-9b, 10a-10b, 11a-11b, 12a-12b and 13a-13b, are the five current waveforms and Cycle Phases Labeled Diagram. Among them, in the periodic stage mark diagram, "-2" represents the weak burst base value stage, "-1" represents the weak burst peak stage, "1" represents the strong burst base value stage, and "2" represents the strong burst peak value stage. It can be seen that in the case of the five current waveforms, most of the marks of the cycle phases are correct.

The above is only a preferred embodiment of the patent of the present invention, but the scope of protection of the patent of the present invention is not limited thereto. Anyone familiar with the technical field within the scope disclosed by the patent of the present invention, according to the scope of the patent of the present invention The technical solution and its invention patent concept are equivalently replaced or changed, such as the threshold swing parameter q, the threshold swing parameter SR of the current local mean sequence, and the current threshold swing parameter groupSR within the group, etc., all belong to the patent of the present invention. protected range.

Claims (9)

1. the self-adapting detecting method in double PMIG welding current wave form cycle stage, adopt the test platform based on industrial computer and arc welding process detector, described arc welding process detector comprises Hall current sensor, it is characterized in that comprising the following steps:

1) utilize Hall current sensor to detect the welding current of a double PMIG welding welding process, and sampled by data collecting card, the current signal I of gained is outputted to industrial computer;

2) according to mean value and the standard deviation thereof of current signal I, estimation local time range;

3) the electric current average of the every bit in calculated current signal I in local time range, thus form the equal value sequence of electric current local;

4) adopt the flash group residing for dual-threshold voltage Cutoff current local equal value sequence every bit and weak impulse train stage, obtain group phased markers sequence;

5) with group's phased markers sequence for foundation, in the scope of each group, adopt dual-threshold voltage to mark phase of the cycles in this group residing for all signaling points.

2. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 1, is characterized in that: step 2) described estimation local time range, specific as follows:

2.1) threshold value is established to swing parameter, by distinguishing the upper threshold value of calculating current, middle threshold value and lower threshold value with following formula (1), formula (2) and formula (3):

uV＝avg+std×q (1)

mV＝avg (2)

dV＝avg-std×q (3)

Wherein, avg is the mean value of current signal I, and std is the standard deviation of current signal I, and q is that threshold value swings parameter, and uV is the upper threshold value of electric current, and mV is the middle threshold value of electric current, and dV is the lower threshold value of electric current;

2.2) respectively with uV, mV and dV tri-threshold value, mean value uP, mP and the dP in 1 ~ 1000 cycle of statistics current signal I;

2.3) intermediate value of uP, mP and dP three is got, as the reference cycle;

2.4) 0.5 ~ 1000 times of the reference cycle is got as local time range.

3. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 2, is characterized in that: the scope that described threshold value swings parameter q is (0,100).

4. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 1, it is characterized in that: step 4) flash group residing for described employing dual-threshold voltage Cutoff current local equal value sequence every bit and weak impulse train stage, specific as follows:

4.1) mean value of the equal value sequence of calculating current local and standard deviation; If the threshold value of the equal value sequence of an electric current local swings parameter, upper threshold value and lower threshold value by distinguishing the equal value sequence of calculating current local with following formula (4) and formula (5):

ULV＝AVG+STD×SR (4)

DLV＝AVG-STD×SR (5)

Wherein, AVG is the mean value of the equal value sequence of electric current local, and STD is the standard deviation of the equal value sequence of electric current local; SR is that the threshold value of the equal value sequence of electric current local swings parameter, and ULV is the upper threshold value of the equal value sequence of electric current local, and DLV is the lower threshold value of the equal value sequence of electric current local;

4.2) if the value of the equal value sequence of electric current local the 1st is more than or equal to ULV, then marking this point is the flash group stage; If the value of the 1st is less than or equal to DLV, then marking this point is the weak impulse train stage; Otherwise namely the value of the 1st is greater than DLV and is less than ULV, marking this point is the weak impulse train stage;

4.3) if the value of the equal value sequence of electric current local i-th is more than or equal to ULV, then mark at i-th for the flash group stage; If the value of i-th is less than or equal to DLV, then marking at i-th is the weak impulse train stage; Otherwise namely the value of i-th is greater than DLV and is less than ULV, inherit the phased markers of the i-th-1 at i-th; Wherein, i >=2.

5. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 4, is characterized in that: the scope of the threshold value swing parameter SR of the equal value sequence of described electric current local is (0,100).

6. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 1, is characterized in that: in step 5) also comprise before and correct each strong and weak impulse train saltus step marginal point in group phased markers sequence, specific as follows:

To any h in group's phased markers sequence strong and weak impulse train saltus step marginal point k, with the value of the kth point of the equal value sequence of electric current local for threshold value, centered by kth point, the saltus step of high-frequency impulse is detected forward and backward in current signal I, find a most contiguous high-frequency impulse trip point u, replace k, correcting is h strong and weak impulse train saltus step marginal point.

7. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 1, is characterized in that: step 5) the described employing dual-threshold voltage mark phase of the cycles, specific as follows:

5.1) in group's phased markers sequence, any one arises from m point, terminate in the flash group of n-th or weak impulse train, calculate [m, n] in scope, the average of current signal I and standard deviation, upper threshold value and lower threshold value by calculating electric current in group with following formula (6), formula (7):

groupULV＝groupAvg+groupStd×groupSR (6)

groupDLV＝groupAvg-groupStd×groupSR (7)

Wherein, groupAvg is the average of current signal I in [m, n] scope, groupStd is the standard deviation of current signal I in [m, n] scope, and groupSR is that the current threshold in group swings parameter, groupULV is the upper threshold value of electric current in group, and groupDLV is the lower threshold value of electric current in group;

5.2) if the value of current signal I m point is more than or equal to groupULV, then marking this point is peak phase in group; If the value of m point is less than or equal to groupDLV, then marking this point is base value stage in group; Otherwise namely the value of m point is greater than groupDLV and is less than groupULV, marking this point is base value stage in group;

5.3) if be flash group residing for current signal I m point, and be the peak phase in group, then the phase of the cycles of this point be labeled as " flash different mountain value "; If flash group and be the base value stage in group, then the phase of the cycles of this point is labeled as " flash group base value "; If weak impulse train and be the peak phase in group, then the phase of the cycles of this point is labeled as " weak pulse rushes different mountain value "; If weak impulse train and be the base value stage in group, then the phase of the cycles of this point is labeled as " weak impulse train base value ";

5.4) if the value of current signal I m+p point is more than or equal to groupULV, then marking m+p point is peak phase in group; If the value of m+p point is less than or equal to groupDLV, then marking m+p point is base value stage in group; Otherwise namely the value of m+p point is greater than groupDLV and is less than groupULV, group's internal labeling of m+p point succession m+p-1 point; Wherein, 1≤p≤n-m;

5.5) if be flash group residing for current signal I m+p point, and be the peak phase in group, then the phase of the cycles of this point be labeled as " flash different mountain value "; If flash group and be the base value stage in group, then the phase of the cycles of this point is labeled as " flash group base value "; If weak impulse train and be the peak phase in group, then the phase of the cycles of this point is labeled as " weak pulse rushes different mountain value "; If weak impulse train and be the base value stage in group, then the phase of the cycles of this point is labeled as " weak impulse train base value ".

8. the self-adapting detecting method in double PMIG welding current wave form cycle stage according to claim 7, is characterized in that: the scope that the current threshold in described group swings parameter groupSR is (0,100).

9. the self-adapting detecting method in the double PMIG welding current wave form cycle stage according to any one of claim 1-8, is characterized in that: described test platform also comprises welding wire conveying mechanism, walking dolly and guide rail and oscillograph; Described arc welding process detector also comprises voltage sensor.