CN103084708B - Method of identifying deviation of welding joint of rotating electric arc gas shielded welding based on rough set - Google Patents
Method of identifying deviation of welding joint of rotating electric arc gas shielded welding based on rough set Download PDFInfo
- Publication number
- CN103084708B CN103084708B CN201310042456.XA CN201310042456A CN103084708B CN 103084708 B CN103084708 B CN 103084708B CN 201310042456 A CN201310042456 A CN 201310042456A CN 103084708 B CN103084708 B CN 103084708B
- Authority
- CN
- China
- Prior art keywords
- welding
- rough set
- matrix
- deviation
- arc
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Arc Welding In General (AREA)
Abstract
The invention discloses a method of identifying deviation of a welding joint of rotating electric arc gas shielded welding based on a rough set. The method utilizes a welding electric arc electric signal sensor, a welding electric arc position sensor, a data collecting card and the like to collect experimental data. Under a rough set modeling mode, the collected experimental data are used for constructing a decision table through a data preprocessing module, and a rough set model is constructed through an attribute reduction module, an attribute value reduction module and a rule reduction module. Under an online deviation prediction mode, the collected experiment data undergo processing by the data preprocessing module and are matched with the rough set model constructed under the rough set modeling mode. In addition, uncertain reasoning is utilized to predict deviation of the welding joint, and the deviation of the welding joint is output through a statistical module. According to the method of identifying deviation of the welding joint of rotating electric arc gas shielded welding based on the rough set, nonparametric modeling of identification of the deviation of the welding joint is achieved, system composition is simple, anti-interference capability is strong, engineering practicality is good, the welding joint deviation can be extracted in real time, and the method can be applied to a corresponding deviation device for real-time deviation.
Description
Technical field
The present invention relates to welding technology field, refer in particular to a kind of rotating the arc gas based on rough set and protect the recognition methods of welding line deviation.The present invention is directed to rotating the arc gas and protect the weld joint tracking of weldering, adopt rough set method butt welded seam deviation to extract, overcome a lot of algorithms in the past and rely on accurate Mathematical Modeling, the shortcoming being easily interfered during application.
Background technology
Rotating the arc gas is protected welding as a kind of efficient solder technology, at cut deal welding field, applied widely, for guaranteeing welding quality, avoid occurring the defects such as sidewall incomplete fusion, need to guarantee in welding process that electric arc is in groove center, need to carry out effectively to follow the tracks of and control in real time, its key is to extract reliably in real time weld seam deviation.
Sensors has following several types at present: mechanical pick-up device, electromagnetic sensor, sonac, welding temperature field sensor, CCD vision sensor, electric arc sensing etc., arc sensor is compared with other addition type sensor as a kind of online real-time sensing device, have simple in structure, cost is low, do not need plus external equipment, accessibility is good, without good characteristics such as delay and fast response times, especially High Speed Rotating Arc Sensor is highly sensitive, become current and even very important research field of soldered sensor from now on, there is powerful vitality and application prospect, the suitable arc welding robot and being similar on the mechanical device of robot of being applied in.
The key of utilizing arc sensor to carry out weld joint tracking is the change detection deviation signal according to arc current or voltage.Deviation signal is known method for distinguishing at present has a variety of, as intelligent control methods such as traditional mathematics modeling method, fuzzy control, neural net model establishings, but arc welding process is a very complicated Stochastic Dynamic Process, effect due to inherent complex mechanism of the interference of noise and droplet transfer etc. itself, makes the processing of signal become very complicated.Along with the development of intelligent mode recognition technology and the continuous expansions of application thereof such as neutral net, SVMs, statistical model identifications, bring new hope also to the THE WELD SEAM TRACKING TECHNOLOGY research based on arc sensor.
When high speed rotating arc gas is protected weldering weldering, deposite metal is subject to sidewall and restrains effect, while is due to the high deposition rate of welding wire, in molten bath, liquid metal is thicker, liquid level inclination angle is larger, its arc shape has special Changing Pattern, existing electric arc method for sensing effect need to be verified again, and the existence of uncertain factors in a large number in welding process, as the accumulation of rear end, molten bath molten iron, the stirring action of electric arc to molten bath, groove machining accuracy is not high, rigging error, the misalignment that workpiece assembling or thermal deformation cause etc., capital butt welded seam track algorithm proposes test, be difficult to this process to carry out accurate mathematical modeling, thereby intelligent modeling and control method come into one's own.
Existing granted patent application number is 2007100342723, the sensor-type automatic weld tracking control method of magnetic control arc, this patent is mainly for TIG welding arc, another application number is 200910025826.2 patents, during for sine swing, arc sensor adopts the method for least square fitting to obtain current waveform corresponding to different deviations, also not for rotating the arc.
Rough set, as a kind of intelligent modeling and control method, can effectively be processed uncertain information, and can effectively from experimental data, remove redundancy, obtains knowledge.Therefore can utilize it, from experimental data, sum up the forecast model of weld seam deviation.
This patent is protected the weld seam deviation intelligent modeling of weldering for rotating the arc gas, in experimental basis, the Changing Pattern of the welding arc signal of telecommunication is further studied, and used for reference the higher deviation recognition methods of existing weld seam deviation recognition methods proposition reliability.
If the method based on intelligent modeling and control can be applied to weld seam deviation identification, the identification of especially three-dimensional deviation, realize the full-automatic high precision of three-dimensional space curve weld seam and follow the tracks of, thereby the automaticity of raising arc welding robot and special plane expands its application.
Summary of the invention
The object of the invention is in order to address the above problem, provide a kind of rotating arc sensor gas based on rough set to protect the recognition methods of welding line deviation, give full play to rough set in the advantage of processing aspect imperfect, uncertain information, improve current mathematical models and be difficult to set up, intelligent modeling method relies on the situation of experience.
In order to achieve the above object, rotating arc sensor gas based on rough set of the present invention protect the recognition methods of welding line deviation based on weld seam deviation recognition system comprise: the formations such as rotating the arc welding torch, welding arc position sensor, the welding arc signal of telecommunication (curtage) sensor, the source of welding current, PC, data collecting card, mode selector, data preprocessing module, rough set model, uncertain reasoning module, statistical module, attribute reduction module, property value yojan module, Rules Reduction module.Welding arc electric signal sensor is Hall current or voltage sensor; Arc position sensor consists of the grating disc of groove type optoelectronic switch and fluting (or through hole), groove type optoelectronic switch is fixed on the not rotating part of rotating the arc welding torch, the rotating part that grating disc is fixed on rotating the arc welding torch is synchronizeed and is rotated with electric arc, while making electric arc rotate to welding anterior position during installation, the light path of groove type optoelectronic switch is just in time passed perforate or the fluting of grating disc, groove type optoelectronic switch output pulse, thus can be by detecting this pulse judgement arc position; The sensor gained welding arc signal of telecommunication and welding arc position signalling access data capture card, data collecting card is connected with PC.In PC, can select online predictive mode and rough set modeling pattern by mode selector.Rough set modeling pattern is that the welding arc signal of telecommunication of all electric arc swing circles of data collecting card collection is obtained to rough set model through data preprocessing module, attribute reduction module, property value yojan module, Rules Reduction resume module successively.This rough set model is used under on-line prediction pattern.Under on-line prediction pattern, using the current electric arc swing circle welding arc signal of telecommunication of data collecting card collection input signal as rough set model after data preprocessing module is processed, and adopt uncertain reasoning resume module, obtain the weld seam deviation of current electric arc swing circle; For overcoming the impact of accidentalia, the weld seam deviation of current electric arc swing circle, with together with n electric arc swing circle weld seam deviation before, as the input of statistical module, is carried out to weld seam deviation output after processing.
Rotating the arc gas based on rough set is protected the recognition methods of welding line deviation, comprises following steps:
(1) adopting Hall element to measure the welding arc signal of telecommunication is welding arc electric current or welding arc voltage; Adopt the grating disc of groove type optoelectronic switch and perforate/fluting to detect the signal welding arc position signalling of rotating the arc when welding the place ahead;
(2) institute's step (1) is detected to the welding arc signal of telecommunication and the welding arc position signalling that obtain and be linked into PC by data collecting card, juxtaposition mode selector is for entering on-line prediction pattern;
(3) in PC, adopt the welding arc signal of telecommunication of the current electric arc swing circle that data preprocessing module obtains data collecting card to carry out pretreatment, obtain the input signal that is applicable to rough set model;
(4), by rough set model input signal process rough set model obtained above and uncertain reasoning module, obtain the weld seam deviation value of current electric arc swing circle;
(5) by the weld seam deviation value of current electric arc swing circle and the weld seam deviation value of n electric arc swing circle before, through statistical module, select weld seam deviation predicted value that arithmetic mean value wherein or weighted mean or the frequency of occurrences are the highest as the final weld seam deviation value output of current electric arc swing circle, wherein n is greater than 1 natural number.
The described data preprocessing module concrete steps of step (3) are as follows:
1) utilize welding arc position signalling to extract the welding arc signal of telecommunication of current electric arc swing circle;
2) the welding arc signal of telecommunication in this electric arc swing circle is carried out to the interval division such as even number, extract the difference of the corresponding interval welding arc signal of telecommunication mean value in each interval welding arc signal of telecommunication mean value and left and right;
3) difference of the corresponding interval welding arc signal of telecommunication mean value of above-mentioned each interval welding arc signal of telecommunication mean value and left and right is carried out to discretization, be about to continuous property and be converted to Category Attributes value;
4) by the one dimension ordered series of numbers output obtaining after step 3) discretization.
The described rough set model of step (4) adopts following steps to obtain:
A) span possible according to weld seam deviation, equidistantly selects the i.e. default weld seam deviation value of limited value, will preset weld seam deviation value and weld respectively experiment, and gather the welding arc signal of telecommunication and welding arc position signalling according to step (1);
B) step (1) is detected to the welding arc signal of telecommunication and the welding arc position signalling that obtain and be linked into PC by data collecting card, juxtaposition mode selector is for entering rough set modeling pattern;
C) in PC, adopt the welding arc signal of telecommunication of all electric arc swing circles that data preprocessing module obtains data collecting card to carry out pretreatment, after pretreatment, the output signal of each electric arc swing circle is inputted as a sample, the default weld seam deviation value that each electric arc swing circle is corresponding is exported as a sample, all compositions of sample rough set decision table, i.e. an input/output list;
D) utilize attribute reduction module to carry out yojan to rough set decision table, remove the input row of redundancy in rough set decision table, i.e. attribute item;
E), for the rough set decision table after step d) processing, adopt property value yojan module to remove the redundant input item in each sample, i.e. property value item in rough set decision table;
F) for step e) data after processing, utilize the minimum rule set that covers set of Rules Reduction module construction, i.e. rough set model.
Wherein, Algorithm for Reduction adopts based on distinguishing matrix algorithm described in step d), and concrete steps are as follows:
I) calculate the differentiation matrix D M of rough set decision table, add up the probability that each conditional attribute occurs, using this as Importance of Attributes;
Ii) extract and distinguish the item that in matrix D M, element is single conditional attribute, ask for its union, obtain the attribute reduction core of rough set decision table, assignment is to property set D;
Iii) common factors all and D are not that empty differentiation matrix element is set to empty set;
Iv) if distinguishing element in matrix is all empty set, D is attribute reduction, termination routine; Otherwise, turn next step;
V) from distinguish matrix, in remaining conditional attribute collection E, according to Importance of Attributes definition, extract most important conditional attribute, assignment is to a, and all with { common factor of a} is not set to empty set for empty differentiation matrix element, and a is added in D, and delete from E, turn the iv) step.
Described in step e), algorithm for attribute value reduction adopts the algorithm based on distinguishing matrix, and concrete steps are as follows:
(I) for gained rough set decision table after step d) attribute reduction, select a unselected sample h, calculate sample h and distinguish matrix D M
h;
(II) calculate in this differentiation matrix by the union F of single conditional attribute component, be the property value yojan core of this sample;
(III) arrange with F and occur simultaneously not for empty differentiation matrix element is empty set, the differentiation matrix of preserving is now
(IV) if DM
hfor empty matrix, the attribute reduction that F is this sample, goes to step (V); Otherwise select an attribute b at random from matrix, and in the sub matrix of setting area with { b} occurs simultaneously for empty element is empty set, goes to step (IV);
(V) if tried to achieve the property value yojan of k this sample, go to step (VI); Otherwise, will
assignment is to DM
h, go to step (IV); K is greater than 1 natural number;
(VI), if all samples of rough set decision table are all selected, the property value yojan of rough set decision table solves end, otherwise, go to step (I).
Described in step f), rule reduction algorithm adopts the algorithm based on distinguishing matrix, and concrete steps are as follows:
A) according to step e) the property value yojan of calculating, calculate it and distinguish matrix D M3, add up the probability of every appearance, using this as Importance of Attributes;
B) extracting element in differentiation matrix D M3 is single item, asks for its union, obtains Rules Reduction core, and assignment is to property set G;
C) common factors all and G are not that empty differentiation matrix element is set to empty set;
D) if distinguishing element in matrix is all empty set, G is Rules Reduction, termination routine; Otherwise, turn next step;
E) from distinguish matrix, in the collection H of residual term, according to Importance of Attributes definition, extract most important, assignment is to c, and all with the common factor of c} is not that empty differentiation matrix element is not set to empty set, and c is added in G, and deletes from H, turns D) step.
Rotating the arc gas based on rough set of the present invention is protected feature and the beneficial effect of the welding deviation recognition methods of weldering:
1, method of the present invention is to have a large amount of uncertain factors for welding process, its essence is to become and the non-linear process of time lag in a high-order, non-linear, time, utilize rough set to process the ability of uncertain information, build nonparametric model, improve the adaptive capacity of model.
2, existing welding seam tracking method is mainly to rely on current/voltage extreme value or integration method.But can causing local extremum, instantaneous short circuit during welding process make extremum method occur erroneous judgement.Integration method or mean value method are just for the certain interval between left and right region, and method of the present invention is to adopt the method for Intelligent treatment to sum up useful information from data itself, can apply better each side information.
Accompanying drawing explanation
Fig. 1 is that the rotating the arc gas based on rough set is protected welding line deviation recognition system schematic diagram;
Fig. 2 is data preprocessing module flow chart;
Fig. 3 is attribute reduction module flow chart in rough set modeling;
Fig. 4 is property value yojan module flow chart in rough set modeling;
Fig. 5 is Rules Reduction module flow chart in rough set modeling.
The specific embodiment
In order to deepen the understanding of the present invention, below in conjunction with the drawings and specific embodiments, the present invention is further described in detail, this embodiment only, for explaining the present invention, does not form limiting the scope of the present invention.
As shown in Figure 1, rotating arc sensor gas based on rough set of the present invention protect the recognition methods of welding line deviation based on weld seam deviation recognition system comprise: rotating the arc welding torch (1), welding arc position sensor (2-1, 2-2), the welding arc signal of telecommunication (curtage) sensor (3), the source of welding current (4), PC (7), data collecting card (8), mode selector (9), data preprocessing module (12), rough set model (13), uncertain reasoning module (14), statistical module (17), attribute reduction module (19), property value yojan module (20), Rules Reduction modules etc. form (21).Welding arc electric signal sensor (3) is Hall current sensor; Arc position sensor (2-1,2-2) grating disc (2-2) by groove type optoelectronic switch (2-1) and fluting (or through hole) forms, groove type optoelectronic switch (2-1) is fixed on the not rotating part of rotating the arc welding torch (1), the rotating part that grating disc (2-2) is fixed on rotating the arc welding torch (1) is synchronizeed and is rotated with electric arc, while making electric arc rotate to welding anterior position during installation, the light path of groove type optoelectronic switch (2-1) is just in time passed perforate or the fluting of grating disc (2-2), groove type optoelectronic switch (2-1) output pulse, thus can be by detecting this pulse judgement arc position; The sensor gained welding arc signal of telecommunication (6) and welding arc position signalling (5) access data capture card (8), data collecting card (8) is connected with PC (7).In PC (7), can select online predictive mode (10) and rough set modeling pattern (11) by mode selector (9).Rough set modeling pattern (11) is that the welding arc signal of telecommunication (6) of all electric arc swing circles that data collecting card (8) is gathered passes through successively data preprocessing module (12), attribute reduction module (19), property value yojan module (20), Rules Reduction module (21) and processes and obtain rough set model (13).This rough set model (13) is used under on-line prediction pattern (10).Under on-line prediction pattern (10), using the current electric arc swing circle welding arc signal of telecommunication (6) of data collecting card (8) collection input signal as rough set model (13) after data preprocessing module (12) is processed, and adopt uncertain reasoning module (14) to process, obtain the weld seam deviation (15) of current electric arc swing circle, and with together with n electric arc swing circle weld seam deviation (16) before, as the input of statistical module (17), after processing, carry out weld seam deviation output (18).
The step of the method is as shown in Figure 1, specific as follows:
(1) adopt the welding arc signal of telecommunication (being electric current) sensor (3) to measure the welding arc signal of telecommunication (6) herein; Adopt the grating disc (2-2) of groove type optoelectronic switch (2-1) and perforate/fluting to detect the signal of rotating the arc when welding the place ahead, i.e. welding arc position signalling (5);
(2) the gathered welding arc signal of telecommunication (6) and welding arc position signalling (5) are linked into PC (7) by data collecting card (8), juxtaposition mode selector (9) is for entering on-line prediction pattern (10);
(3) in PC, adopt the current electric arc swing circle welding arc signal of telecommunication (6) that data preprocessing module (12) is obtained data collecting card (8) to carry out pretreatment, obtain the input signal that is applicable to rough set model (13);
(4) by above-mentioned rough set model (13) input signal that obtains, utilize rough set model (13) and uncertain reasoning module (14), obtain the weld seam deviation value (15) of current electric arc swing circle; Wherein uncertain reasoning module (14) adopts rule match in rough set model (13) input signal and rough set model (13), and 5 rules that mate most of selecting matching degree to be greater than 60% are got the weighted average of its predicted value;
(5) consider current electric arc swing circle weld seam deviation value (15) and n electric arc swing circle weld seam deviation value (16) before, through statistical module (17), select the weld seam deviation predicted value that its arithmetic mean value or weighted mean or the frequency of occurrences are the highest to export (18) as the final weld seam deviation value of current electric arc swing circle.N value is 10 herein, and statistical module (17) calculates arithmetic mean of instantaneous value.
The described data preprocessing module of step (3) (12) algorithm flow is as shown in Figure 2, specific as follows:
1) utilize welding arc position signalling (5) to extract the welding arc signal of telecommunication (6) of current electric arc swing circle;
2) the welding arc signal of telecommunication (6) in this electric arc swing circle is carried out to the interval division such as even number, extract the difference of the corresponding interval welding arc signal of telecommunication (6) mean value in each the interval welding arc signal of telecommunication (6) mean value and left and right; Even number is taken as 12 herein;
3) difference of the corresponding interval welding arc signal of telecommunication (6) mean value of above-mentioned each the interval welding arc signal of telecommunication (6) mean value and left and right is carried out to discretization, be about to continuous property and be converted to Category Attributes value; Adopt the discretization algorithm based on entropy to carry out discretization to the input in rough set decision table herein, and set its stop condition according to itself and the coefficient correlation size of output; When coefficient correlation absolute value being set herein belonging to [0,0.3], maximum distributing breakpoint number is 2, coefficient correlation absolute value belong to (0.3,0.6] time maximum distributing breakpoint number be 3, coefficient correlation absolute value belong to (0.6,1] time maximum distributing breakpoint number be 4;
4) by the one dimension ordered series of numbers output after discretization.
The described rough set model obtaining step of step (4) is as shown in Figure 1, specific as follows:
A) span possible according to weld seam deviation, limited value of equidistant selection, default these weld seam deviation values are welded respectively experiment, be respectively+1mm of weld seam deviation ,+0.5mm, 0mm ,-0.5mm ,-1mm are set herein, wherein positive number is left avertence, negative is right avertence, and gathers the welding arc signal of telecommunication (6) and welding arc position signalling (5) according to step (1);
B) the gathered welding arc signal of telecommunication (6) and welding arc position signalling (5) are linked into PC (7) by data collecting card (8), juxtaposition mode selector (9) is for entering rough set modeling pattern (11);
C) all electric arc swing circle welding arc signals of telecommunication (6) that adopt data preprocessing module (12) to obtain data collecting card (8) in PC (7) carry out pretreatment, after pretreatment, the output signal of each electric arc swing circle is inputted as a sample, the default weld seam deviation value that each electric arc swing circle is corresponding is exported as a sample, all compositions of sample rough set decision table, i.e. an input/output list;
D) utilize attribute reduction module (19) to carry out yojan to rough set decision table, remove the input row of redundancy in rough set decision table, i.e. attribute item.
Algorithm for Reduction adopts based on distinguishing matrix algorithm, concrete steps as shown in Figure 3:
I) calculate the differentiation matrix D M of rough set decision table, add up the probability that each conditional attribute occurs, using this as Importance of Attributes;
Ii) extract and distinguish the item that in matrix D M, element is single conditional attribute, ask for its union, obtain the attribute reduction core of rough set decision table, assignment is to property set D;
Iii) common factors all and D are not that empty differentiation matrix element is set to empty set;
Iv) if distinguishing element in matrix is all empty set, D is attribute reduction, termination routine; Otherwise, turn next step;
V) from distinguish matrix, in remaining conditional attribute collection E, according to Importance of Attributes definition, extract most important conditional attribute, assignment is to a, and all with { common factor of a} is not set to empty set for empty differentiation matrix element, and a is added in D, and delete from E, turn the iv) step;
E), for the rough set decision table after step d) processing, adopt property value yojan module (20) to remove the redundant input item in each sample, i.e. property value item in rough set decision table.Algorithm for attribute value reduction adopts the algorithm based on distinguishing matrix, concrete steps as shown in Figure 4:
(I) for gained rough set decision table after step d) attribute reduction, select a unselected sample h, calculate sample h and distinguish matrix D M
h;
(II) calculate in this differentiation matrix by the union F of single conditional attribute component, be the property value yojan core of this sample;
(III) setting and F common factor are not that empty differentiation matrix element is sky, and preservation differentiation matrix is now
(IV) if DM
hfor empty matrix, the attribute reduction that F is this sample, goes to step (V); Otherwise select an attribute b at random from matrix, and in the sub matrix of setting area with { b} occurs simultaneously for empty element is empty set, goes to step (IV);
(V) if tried to achieve the property value yojan of k this sample, go to step (VI); Otherwise, will
assignment is to DM
h, go to step (IV); K value is 5 herein.
(VI), if all samples of rough set decision table are all selected, the property value yojan of rough set decision table solves end, otherwise, go to step (I);
F) for step e) data after processing, utilize Rules Reduction module (21) to build the minimum rule set that covers set, i.e. rough set model (13).
Rule reduction algorithm adopts the algorithm based on distinguishing matrix, concrete steps as shown in Figure 5:
A) according to step e) the property value yojan of calculating, calculate it and distinguish matrix D M3, add up the probability of every appearance, using this as Importance of Attributes;
B) extracting element in differentiation matrix D M3 is single item, asks for its union, obtains Rules Reduction core, and assignment is to property set G;
C) common factors all and G are not that empty differentiation matrix element is set to empty set;
D) if distinguishing element in matrix is all empty set, G is Rules Reduction, termination routine; Otherwise, turn next step;
E) from distinguish matrix, in the collection H of residual term, according to Importance of Attributes definition, extract most important, assignment is to c, and all with the common factor of c} is not that empty differentiation matrix element is not set to empty set, and c is added in G, and deletes from H, turns D) step.
Claims (4)
1. the rotating the arc gas based on rough set is protected the recognition methods of welding line deviation, comprises following steps:
(1) adopt Hall element to measure the welding arc signal of telecommunication, i.e. welding arc electric current or welding arc voltage; Adopt the grating disc of groove type optoelectronic switch and perforate/fluting to detect the signal welding arc position signalling of rotating the arc when welding the place ahead;
(2) step (1) is detected to the welding arc signal of telecommunication and the welding arc position signalling that obtain and be linked into PC by data collecting card, juxtaposition mode selector is for entering on-line prediction pattern;
(3) in PC, adopt the welding arc signal of telecommunication of the current electric arc swing circle that data preprocessing module obtains data collecting card to carry out pretreatment, obtain the input signal that is applicable to rough set model;
(4), by rough set model input signal process rough set model obtained above and uncertain reasoning module, obtain the weld seam deviation value of current electric arc swing circle;
(5) by the weld seam deviation value of current electric arc swing circle and the weld seam deviation value of n electric arc swing circle before, through statistical module, select weld seam deviation predicted value that arithmetic mean value wherein or weighted mean or the frequency of occurrences are the highest as the final weld seam deviation value output of current electric arc swing circle, wherein n is greater than 1 natural number;
It is characterized in that the described data preprocessing module concrete steps of step (3) are as follows:
1) utilize welding arc position signalling to extract the welding arc signal of telecommunication of current electric arc swing circle;
2) the welding arc signal of telecommunication in this electric arc swing circle is carried out to the interval division such as even number, extract the difference of the corresponding interval welding arc signal of telecommunication mean value in each interval welding arc signal of telecommunication mean value and left and right;
3) difference of the corresponding interval welding arc signal of telecommunication mean value of above-mentioned each interval welding arc signal of telecommunication mean value and left and right is carried out to discretization, be about to continuous property and be converted to Category Attributes value;
4) by step 3) the one dimension ordered series of numbers output that obtains after discretization;
The described rough set model of step (4) adopts following steps to obtain:
A) span possible according to weld seam deviation, equidistantly selects the i.e. default weld seam deviation value of limited value, will preset weld seam deviation value and weld respectively experiment, and gather the welding arc signal of telecommunication and welding arc position signalling according to step (1);
B) step (1) is detected to the welding arc signal of telecommunication and the welding arc position signalling that obtain and be linked into PC by data collecting card, juxtaposition mode selector is for entering rough set modeling pattern;
C) in PC, adopt the welding arc signal of telecommunication of all electric arc swing circles that data preprocessing module obtains data collecting card to carry out pretreatment, after pretreatment, the output signal of each electric arc swing circle is inputted as a sample, the default weld seam deviation value that each electric arc swing circle is corresponding is exported as a sample, all compositions of sample rough set decision table, i.e. an input/output list;
D) utilize attribute reduction module to carry out yojan to rough set decision table, remove the input row of redundancy in rough set decision table, i.e. attribute item;
E) for steps d) rough set decision table after processing, adopt property value yojan module to remove the redundant input item in each sample, i.e. property value item in rough set decision table;
F) for step e) data after processing, utilize the minimum rule set that covers set of Rules Reduction module construction, i.e. rough set model.
2. rotating the arc gas based on rough set according to claim 1 is protected the recognition methods of welding line deviation, it is characterized in that steps d) algorithm of described yojan is to adopt based on distinguishing matrix algorithm, concrete steps are as follows:
I) calculate the differentiation matrix D M of rough set decision table, add up the probability that each conditional attribute occurs, using this as Importance of Attributes;
Ii) extract and distinguish the item that in matrix D M, element is single conditional attribute, ask for its union, obtain the attribute reduction core of rough set decision table, assignment is to property set D;
Iii) common factors all and D are not that empty differentiation matrix element is set to empty set;
Iv) if distinguishing element in matrix is all empty set, D is attribute reduction, termination routine; Otherwise, turn next step;
V) from distinguish matrix, in remaining conditional attribute collection E, according to Importance of Attributes definition, extract most important conditional attribute, assignment is to a, and all with { common factor of a} is not set to empty set for empty differentiation matrix element, and a is added in D, and delete from E, turn iv) step.
3. rotating the arc gas based on rough set according to claim 1 is protected the recognition methods of welding line deviation, it is characterized in that step e) algorithm of described property value yojan is the algorithm adopting based on distinguishing matrix, concrete steps are as follows:
(I) for steps d) gained rough set decision table after attribute reduction, select a unselected sample h, calculate sample h and distinguish matrix D M
h;
(II) calculate in this differentiation matrix by the union F of single conditional attribute component, be the property value yojan core of this sample;
(III) arrange with F and occur simultaneously not for empty differentiation matrix element is empty set, the differentiation matrix of preserving is now DM
h 2;
(IV) if DM
hfor empty matrix, the attribute reduction that F is this sample, goes to step (V); Otherwise select an attribute b at random from matrix, and in the sub matrix of setting area with { b} occurs simultaneously for empty element is empty set, goes to step (IV);
(V) if tried to achieve the property value yojan of k this sample, go to step (VI); Otherwise, by DM
h 2assignment is to DM
h, go to step (IV); K is greater than 1 natural number;
(VI), if all samples of rough set decision table are all selected, the property value yojan of rough set decision table solves end, otherwise, go to step (I).
4. rotating the arc gas based on rough set according to claim 1 is protected the recognition methods of welding line deviation, it is characterized in that step f) algorithm of described Rules Reduction is the algorithm adopting based on distinguishing matrix, concrete steps are as follows:
A) according to step e) the property value yojan of calculating, calculate it and distinguish matrix D M3, add up the probability of every appearance, using this as Importance of Attributes;
B) extracting element in differentiation matrix D M3 is single item, asks for its union, obtains Rules Reduction core, and assignment is to property set G;
C) common factors all and G are not that empty differentiation matrix element is set to empty set;
D) if distinguishing element in matrix is all empty set, G is Rules Reduction, termination routine; Otherwise, turn next step;
E) from distinguish matrix, in the collection H of residual term, according to Importance of Attributes definition, extract most important, assignment is to c, and all with the common factor of c} is not that empty differentiation matrix element is not set to empty set, and c is added in G, and deletes from H, turns D) step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310042456.XA CN103084708B (en) | 2013-02-04 | 2013-02-04 | Method of identifying deviation of welding joint of rotating electric arc gas shielded welding based on rough set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310042456.XA CN103084708B (en) | 2013-02-04 | 2013-02-04 | Method of identifying deviation of welding joint of rotating electric arc gas shielded welding based on rough set |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103084708A CN103084708A (en) | 2013-05-08 |
| CN103084708B true CN103084708B (en) | 2014-11-05 |
Family
ID=48198110
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310042456.XA Expired - Fee Related CN103084708B (en) | 2013-02-04 | 2013-02-04 | Method of identifying deviation of welding joint of rotating electric arc gas shielded welding based on rough set |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103084708B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103464869B (en) * | 2013-08-26 | 2016-01-13 | 江苏科技大学 | The rotating the arc narrow gap MAG welding line deviation recognition device of view-based access control model sensing and method |
| CN103639568A (en) * | 2013-12-25 | 2014-03-19 | 江苏新瑞峰信息科技有限公司 | Welding arc rocking device |
| CN103760456B (en) * | 2013-12-31 | 2016-08-17 | 合肥天海电气技术有限公司 | Electrical network capacitance type equipment insulation based on rough set theory monitoring system |
| CN104070264B (en) * | 2014-06-23 | 2016-01-13 | 江苏科技大学 | The narrow gap MAG of rotating the arc becoming groove width welds self adaptation pendulous device and method |
| JP6776493B2 (en) * | 2016-11-16 | 2020-10-28 | 株式会社神戸製鋼所 | Misalignment detection method in arc copying welding |
| CN107570841B (en) * | 2017-10-31 | 2019-03-19 | 昆山华恒焊接股份有限公司 | The control method and control device of arc energy, computer-readable medium |
| CN109813728A (en) * | 2019-03-01 | 2019-05-28 | 沈阳建筑大学 | A kind of circuit board solder joint detection method and system |
| CN110571802B (en) * | 2019-09-12 | 2023-03-28 | 国网山东省电力公司寿光市供电公司 | Power distribution network self-healing control method and device based on rough set dynamic planning model |
| CN112765751B (en) * | 2021-01-27 | 2022-04-19 | 南昌航空大学 | Machining deformation control method for aluminum alloy thick plate in milling process |
| CN113240095B (en) * | 2021-06-07 | 2022-04-22 | 北京理工大学 | Casting cylinder cover mechanical property prediction method based on rough set and neural network |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1546288A (en) * | 2003-12-17 | 2004-11-17 | 南昌大学 | Curved weld seam independent traveling welding robot system |
| CN101514886A (en) * | 2009-03-10 | 2009-08-26 | 东南大学 | Method for extracting arc sensor welding gun position deviation information |
| CN102615390A (en) * | 2012-04-28 | 2012-08-01 | 湘潭大学 | Swing arc-based multi-layer and multi-channel weld tracking system and identification method thereof |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009166076A (en) * | 2008-01-15 | 2009-07-30 | Kobe Steel Ltd | Welding robot |
-
2013
- 2013-02-04 CN CN201310042456.XA patent/CN103084708B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1546288A (en) * | 2003-12-17 | 2004-11-17 | 南昌大学 | Curved weld seam independent traveling welding robot system |
| CN101514886A (en) * | 2009-03-10 | 2009-08-26 | 东南大学 | Method for extracting arc sensor welding gun position deviation information |
| CN102615390A (en) * | 2012-04-28 | 2012-08-01 | 湘潭大学 | Swing arc-based multi-layer and multi-channel weld tracking system and identification method thereof |
Non-Patent Citations (6)
| Title |
|---|
| "基于变精度粗糙集的脉冲GTAW过程建模方法";黎文航;《焊接学报》;20080731;第29卷(第7期);第57-63页 * |
| "基于粗糙集理论的焊接过程建模方法";王兵等;《控制理论与应用》;20040630;第21卷(第3期);第411-414页 * |
| 朱杰等."焊接电弧摇动定位控制系统研究".《江苏科技大学学报（自然科学版）》.2012,第26卷(第5期),第448-451页. * |
| 朱杰等."焊接电弧摇动定位控制系统研究".《江苏科技大学学报(自然科学版)》.2012,第26卷(第5期),第448-451页. * |
| 王兵等."基于粗糙集理论的焊接过程建模方法".《控制理论与应用》.2004,第21卷(第3期),第411-414页. * |
| 黎文航."基于变精度粗糙集的脉冲GTAW过程建模方法".《焊接学报》.2008,第29卷(第7期),第57-63页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103084708A (en) | 2013-05-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103084708B (en) | Method of identifying deviation of welding joint of rotating electric arc gas shielded welding based on rough set | |
| CN105118044B (en) | A kind of wheel shape cast article defect automatic testing method | |
| CN104390657B (en) | A kind of Generator Unit Operating Parameters measurement sensor fault diagnosis method and system | |
| CN109318055B (en) | Milling cutter wear state feature extraction multi-objective optimization method | |
| Liang et al. | Defect detection of rail surface with deep convolutional neural networks | |
| CN106338521A (en) | Additive manufacturing surface defect, internal defect and shape composite detection method and device | |
| CN109977916B (en) | Online monitoring system for dissolved gas in transformer oil based on embedded platform | |
| CN108971704B (en) | Three-dimensional weld tracking method and system based on vision and rotating arc composite sensing | |
| CN103357987B (en) | Automatic stability detecting method for process of CO2 electric arc welding short circuit transition welding | |
| CN103143815B (en) | Weld joint deviation identification method, based on waveform matching, of rotating arc narrow gap gas shield welding | |
| CN107627152A (en) | A kind of digital control processing chip control method based on BP neural network | |
| CN108956041B (en) | Sintering machine air leakage fault online diagnosis method based on single-side detection method of frequency spectrum characteristics | |
| CN115358155A (en) | Power big data abnormity early warning method, device, equipment and readable storage medium | |
| Tang et al. | An online surface defects detection system for AWAM based on deep learning | |
| CN104181196A (en) | On-line detection method of continuous casting billet surface longitudinal crack | |
| Ak et al. | Data analytics and uncertainty quantification for energy prediction in manufacturing | |
| Yang et al. | Detection of weld groove edge based on multilayer convolution neural network | |
| CN110597065B (en) | Adaptive control system for powder feeding type laser additive | |
| CN103179332B (en) | A kind of sensation target self-adapting detecting controller based on FPGA technology | |
| CN104460514A (en) | Method for fusion predicting on machine tool size errors through information | |
| CN117572863A (en) | Path optimization method and system for substation robot | |
| CN110918973B (en) | Crystallizer thermal image abnormal region marking method based on run | |
| CN113487149B (en) | Welding spot abnormity identification system and method based on Catboost K-fold cross verification | |
| CN113554869B (en) | Road closure detection method based on multi-feature fusion | |
| CN111438460B (en) | Vision-based thick plate T-shaped joint welding seam forming characteristic online measurement method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20141105 Termination date: 20170204 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |