CN113505657A - Welding spot quality detection method and device - Google Patents
Welding spot quality detection method and device Download PDFInfo
- Publication number
- CN113505657A CN113505657A CN202110679149.7A CN202110679149A CN113505657A CN 113505657 A CN113505657 A CN 113505657A CN 202110679149 A CN202110679149 A CN 202110679149A CN 113505657 A CN113505657 A CN 113505657A
- Authority
- CN
- China
- Prior art keywords
- welding spot
- welding
- resistance
- spot
- points
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000003466 welding Methods 0.000 title claims abstract description 371
- 238000001514 detection method Methods 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 claims abstract description 113
- 230000008569 process Effects 0.000 claims abstract description 78
- 238000013145 classification model Methods 0.000 claims abstract description 43
- 238000012549 training Methods 0.000 claims abstract description 25
- 229910000679 solder Inorganic materials 0.000 claims description 25
- 238000007689 inspection Methods 0.000 claims description 9
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 5
- 230000035772 mutation Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000009795 derivation Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 2
- 238000010801 machine learning Methods 0.000 abstract description 18
- 230000008859 change Effects 0.000 description 9
- 230000006872 improvement Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 7
- 239000002184 metal Substances 0.000 description 6
- 238000011160 research Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 239000007769 metal material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005476 soldering Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Resistance Welding (AREA)
Abstract
The invention discloses a method and a device for detecting the quality of a welding spot. The method comprises the steps of obtaining technological parameters of a plurality of welding points in the welding process; constructing a resistance time relation curve of each welding point according to the process parameters; extracting the characteristic value of each welding spot from the resistance-time relation curve to obtain a welding spot characteristic value set; training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states; and judging the welding spot state of the welding spot to be detected by utilizing the classification model. The method is based on a big data mathematical statistics analysis means, analyzes a large amount of accumulated welding process data which are automatically collected in real time, performs model training by adopting a TPOT automatic machine learning algorithm, further quickly and efficiently judges the mass welding spot states by utilizing a classification model, reduces equipment investment of welding spot detection, reduces cost, and provides reliable data for welding spot performance and welding spot process quality.
Description
Technical Field
The invention relates to the field of solder joint quality detection, in particular to a solder joint quality detection method and a solder joint quality detection device.
Background
Welding is a common process in modern machine manufacturing and is also widely used in automotive manufacturing. The welding process takes a welding gun as a tool, two or more than two kinds of same or different metal materials are connected into a whole in a welding spot mode, and the quality of the welding spot has great influence on the overall quality of the automobile. In the existing technical scheme, the judgment of the welding spot state mainly includes that welding spot data and welding spot splashing posture photos are collected on line and researched by installing a special device and a sensor.
At present, the research on industrial welding spot big data in the industry is less, the research is mainly focused on the aspects of welding spot splashing, energy consumption and the like, and various states of the welding spot are identified by a big data machine learning means, particularly an automatic machine learning algorithm, or the welding spot is blank.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method and a device for detecting the quality of a welding spot, which can efficiently and accurately complete the research on a large amount of welding spot information, judge the state of the welding spot, provide data support for further analysis of a welding process, do not need to install equipment and have low cost.
In order to solve the technical problem, the welding spot quality detection method comprises the following steps: acquiring technological parameters of a plurality of welding points in a welding process, wherein the technological parameters comprise the numbers, currents, voltages, resistances and powers of the welding points and corresponding welding time points; constructing a resistance time relation curve of each welding point according to the process parameters; extracting the characteristic value of each welding spot from the resistance time relation curve to obtain a welding spot characteristic value set; training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar points; and judging the welding spot state of the welding spot to be detected by utilizing the classification model.
In the welding spot quality detection method, the resistance time relation curve of each target welding spot is obtained by analyzing and processing the process parameter information of mass welding spots, the characteristic value set is further extracted, and a classification model between the welding spot state and the characteristic value set is trained by machine learning, so that the welding spot data can be rapidly and efficiently processed by using the classification model, the welding spot state is detected, no special equipment is invested, and the cost is low.
As an improvement of the welding spot quality detection method, the characteristic value of each welding spot is extracted from the resistance-time relation curve, and the obtained welding spot characteristic value set comprises the following steps: extracting a resistance minimum value in a first time interval; extracting the maximum value of the resistance in the second time interval; extracting the resistance mutation points in the third time interval, and counting to obtain the number of the splash points; and forming a welding spot characteristic value set by the extracted resistance minimum value, the resistance maximum value and the number of the spattered points. The change characteristics of the resistance of the welding spot in the welding process are utilized to represent the state of the welding spot, and because the resistance data is easy to obtain and analyze, the forming process of the welding spot can be better reflected, and the rapid and accurate judgment on the state of the welding spot is further realized.
In the method for detecting the quality of the welding spots, the first time interval is 0-10ms from the beginning of welding of each welding spot, the second time interval is 20-30ms from the beginning of welding of each welding spot, and the third time interval is 30ms from the beginning to the end of welding of each welding spot. The three time intervals correspond to three phases of the welding process, respectively.
For further improvement of the welding spot quality detection method, training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states comprises the following steps: if the number of the spattering points exceeds one, outputting that the welding spot state is multipoint spattering; if the number of the spattering points is one, outputting that the welding spot state is single-point spattering; if the number of the splashing points is zero and the resistance minimum value is greater than or equal to the resistance maximum value, outputting a welding spot state as an electrodeless value point; and if the number of the splashing points is zero and the resistance minimum value is smaller than the resistance maximum value, outputting that the welding spot state is normal.
For further improvement of the welding spot quality detection method, the step of judging the welding spot state of the welding spot to be detected by using the classification model comprises the following steps: acquiring process parameters of a welding spot to be detected in a welding process; constructing a resistance-time relation curve of the welding spot to be detected according to the technological parameters of the welding spot to be detected in the welding process; extracting a resistance minimum value, a resistance maximum value and the number of splash points from a resistance time relation curve of a welding spot to be detected to form a welding spot characteristic value set; inputting the welding spot characteristic value set of the welding spot to be detected into the classification model, and according to the characteristic value: and judging the state of the welding spot to which the welding spot to be detected belongs according to the resistance minimum value, the resistance maximum value and the number of the splashing points.
As another improvement of the method for detecting the quality of the welding spot, the method for training the characteristic value set of the welding spot of each welding spot and the state of the welding spot to which each welding spot belongs to obtain a classification model for classifying the states of the welding spots comprises the following specific steps: and optimizing the training flow of the characteristic value set by using a TPOT (tire pressure test) method to obtain a classification model for classifying the states of the welding spots. Model training verification is carried out by adopting a TPOT automatic machine learning algorithm, thousands of possible machine learning pipelines are intelligently explored at the same time, the optimal machine learning pipeline and automatic super-parameter adjustment are obtained, a Python file of the optimal machine learning training pipeline code is automatically generated, the code file can be directly used for retraining a data source, a prediction result and an evaluation index are automatically generated at the same time, compared with the traditional machine learning, the time and energy required by selection of a large number of machine learning models of an algorithm engineer, super-parameter optimization and the like are reduced, and the final result is satisfactory
As another improvement of the method for detecting the quality of the welding spots, after acquiring the process parameters of the plurality of welding spots in the welding process, the method further comprises: and removing the technological parameters of the corresponding welding spots formed when the welding spot fails in the welding process and the welding gun is powered off. Preprocessing the data, deleting meaningless data in the process parameters, ensuring the effectiveness of obtaining the process parameters and improving the analysis processing speed.
As another improvement of the method for detecting quality of a welding spot, after the classification model is used to judge the state of the welding spot, the method further comprises: after the classification model is used for judging the welding spot state of the welding spot to be detected, the method also comprises the following steps: and outputting the process parameters and the welding spot state of the welding spot under the condition that the judgment result of the welding spot state of the welding spot to be detected is abnormal. After the classification model is used for accurately predicting the state of the welding spot, the process parameters of the abnormal welding spot, the welding gun, the electrode cap and the like are further researched, the quality defect problem is quickly found, main factors influencing the quality abnormity are searched, and the predictive maintenance of quality inspection, process and the like is realized, so that the aims of increasing efficiency, saving energy and reducing cost are fulfilled.
In order to solve the technical problem, the welding spot quality detection device comprises an acquisition module, a detection module and a control module, wherein the acquisition module is used for acquiring process parameters of a plurality of welding spots in a welding process, and the process parameters comprise the numbers, the currents, the voltages, the resistances and the powers of the welding spots and corresponding welding time points; the construction module is used for constructing a resistance time relation curve of each welding point according to the process parameters; the extraction module is used for extracting the characteristic value of each welding point from the resistance time relation curve to obtain a welding point characteristic value set; the training module is used for training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar points; and the judging module is used for judging the welding spot state of the welding spot to be detected by utilizing the classification model.
In the welding spot quality detection device, the resistance time curve of the target welding spot is constructed through the construction module according to the technological parameters of the plurality of welding spots, the characteristic values of the welding spots are extracted through the extraction module to form the characteristic value set of the welding spots, the classification model is trained through the training module, so that machine learning is applied to judgment of the welding spot state, the processing speed of the technological parameters of the welding spots is improved, meanwhile, high accuracy is achieved, special equipment does not need to be invested, and the cost is saved.
In conclusion, by adopting the welding spot quality detection method and the welding spot quality detection device, the welding spot state classification model is obtained by the TPOT automatic machine learning method, the classification model is utilized to accurately and rapidly judge the state of the welding spot in the welding process of the welding machine, the detection of the welding spot state is rapidly and accurately finished, the investment of other monitoring equipment is not needed, the cost is saved, in addition, the judgment result is combined with the sampling quality inspection analysis of a welding spot object, the factors influencing the quality and the performance of the welding spot can be conveniently found, the predictive maintenance of quality inspection, process and the like is realized, and the purposes of efficiency improvement, energy conservation and cost reduction are achieved. .
Drawings
In the drawings:
FIG. 1 is a flow chart of a solder joint quality inspection method according to the present invention.
Fig. 2 is a schematic structural diagram of the welding spot quality detection apparatus of the present invention.
FIG. 3 is a graph of resistance versus time for the present invention.
FIG. 4 is a process flow of the solder joint status label of the present invention.
Detailed Description
The following further describes embodiments of the present invention with reference to the drawings. It should be noted that the description of the embodiments is provided to help understanding of the present invention, but the present invention is not limited thereto.
Fig. 1 is a flowchart of a solder joint quality inspection method according to the present invention, and as shown in fig. 1, the solder joint quality inspection method includes the following steps:
step S101: and acquiring process parameters of a plurality of welding points in the welding process.
The influence factors of the welding spot state are reflected on the collected welding machine data, mainly including current, voltage, resistance and power, and corresponding welding time points.
Optionally, after acquiring process parameters of the plurality of welding points in the welding process, the method further includes: and removing the technological parameters of the corresponding welding spots formed when the welding spot fails in the welding process and the welding gun is powered off.
Firstly, cleaning mass data, deleting welding point data files with welding point data time being too long (more than or equal to 400ms) or too short (less than or equal to 200ms), wherein the data is data of welding failure, deleting the welding point data files with current, voltage, resistance and power being 0 in the middle of the welding point data files, and the data is welding gun power failure caused by an emergency in the welding process; and then processing the process parameters of each residual welding spot to ensure the validity of data.
Step S102: and constructing a resistance time relation curve of each welding point according to the process parameters.
In the welding process, the changes of various process parameters are mutually related and can be related with the state of the welding spot, and the research on the state of the welding spot can be completed by selecting a parameter change rule with reference value, namely a characteristic time curve.
The method selects a resistance-time relation curve, extracts voltage and current from process parameters, and generates a resistance-time curve based on the voltage-time curve and the current-time curve.
To describe the relationship between the resistance and the state of the solder joint, the process of soldering is explained as follows:
a complete welding process typically includes: the metal bonding is compact (stage 1), the metal is heated and melted (stage 2), and the welding of the nugget is completed (stage 3), and the resistance-time relation curve of the welding process is shown in figure 3: at the end of the metal bonding and compacting process, along with the change of the metal bonding condition, a resistance value of a resistance curve is a relatively low point R1, and the corresponding time is t 1; after entering a metal heating and melting stage, along with the temperature rise among metal plates, a resistance curve has a relatively high resistance point R2 with corresponding time t2, and then entering a weld nugget forming stage, wherein the corresponding resistance curve is gradually reduced from the high point R2; when the welding current is 0, the welding is finished, the corresponding time is t3, and the corresponding resistance value of the resistance curve is R3.
The abnormal fluctuation (unsmooth) of the R curve of the resistance in fig. 3 indicates that the "spattering" phenomenon occurs in the current welding process, and the larger the fluctuation jump, the more serious the "spattering" phenomenon is, thereby causing the poor welding spot quality and the higher energy consumption in use, and the related comments of the parameters are shown in table 1 below:
TABLE 1 welding procedure-related parameters Table
From the above analysis, it can be known that the judgment of the welding spot state can be completed by analyzing the resistance value time curve. It should be noted that, in the present application, specific solder joint features are not limited, and only defining features is helpful for solder joint state classification.
Step S103: and extracting the characteristic value of each welding point from the resistance-time relation curve to obtain a welding point characteristic value set.
Because the resistance-time relationship curve cannot be directly used for identifying the type of the welding spot, the characteristic value of the welding spot needs to be extracted from the characteristic-time curve to obtain the characteristic value set of the welding spot.
Optionally, extracting the characteristic value of each solder joint from the resistance-time relationship curve to obtain a solder joint characteristic value set includes: extracting a resistance minimum value in a first time interval; extracting the maximum value of the resistance in the second time interval; extracting the resistance mutation points in the third time interval, and counting to obtain the number of the splash points; and forming a welding spot characteristic value set by the extracted resistance minimum value, resistance maximum value and the number of the spattered points. From the resistance-time relationship curve of the welding process in fig. 3, corresponding characteristic values are extracted in three stages respectively as basic characteristic parameters for characterizing the state of the welding spot.
Since the relatively low point of the welding dynamic resistance of phase 1 occurs around 5ms of welding, the relatively high point of the welding dynamic resistance of phase 2 occurs around 20ms of welding, and the end of phase 3 welding is around 250 ms. If abnormity exists in the welding process (the gap between two plates is too large and the like), the welding robot can carry out self-adaptive control to adjust current and voltage, the welding time is prolonged, and the whole welding time is increased but cannot exceed 400 ms.
For the characteristic value of each extracted welding point, the first time interval is 0-10ms from the beginning of welding of each welding point, and a relatively low dynamic resistance point which appears within 10ms of welding, namely R1 is taken as a characteristic value of a minimum resistance value; the second time interval is that the dynamic resistance relatively high point appears in the range of 20ms to 30ms from the beginning of welding of each welding point, namely R2 is taken as a resistance maximum value characteristic value; the third time interval is the time from 30ms after welding to the end of welding of each welding point, in order to simplify the complexity of data processing, the judgment of the spatter mainly refers to the spatter in the stage 3, the spatter in the stage 2 is not considered, the point of abnormal fluctuation of the slope of the resistance-time curve in the stage 3 is a resistance abrupt change point, and the number of the resistance abrupt change points, namely the corresponding spatter times, is represented by an include _ num _ T3.
For the extracting of the resistance mutation points in the third time interval, counting the number of the spattering points comprises: carrying out derivation on the resistance time relation curve to obtain a first derivative; acquiring a point which cannot be derived in a third time interval or a point at which the absolute value of the first derivative exceeds a set value; and counting the total number of points in which the absolute values of the derivative points and the first derivative exceed the set value, so as to obtain the number of the splash points.
The specific analysis processing process for extracting the characteristic value may be: firstly, first-order derivation is carried out to obtain a change rate function of the resistance; then, extracting a welding time point influencing the judgment of the welding spot state by utilizing a first-order derivative, wherein the point with the first-order derivative of zero within 10ms is a resistance minimum characteristic value R1, the point with the first-order derivative of zero within 20ms-30ms is a resistance maximum characteristic value R2, and after 30ms, the point with the sudden change is subjected to derivative operation or the change rate exceeds a set value due to the sudden change of the resistance value during splashing, finding out the points and counting the number of the extremely splashed points, namely inflect _ num _ T3; the set of solder joint resistance values (R1, R2, include _ num _ T3) is formed.
Step S104: and training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar value points.
Specifically, as shown in fig. 4, the process of determining the solder joint state by the feature quantities (R1, R2, include _ num _ T3) is as follows: firstly, judging the resistance value quantity contained in the characteristic quantity inflect _ num _ T3, wherein when the inflect _ num _ T3 is greater than 1, welding spots are splashed at multiple points;
when the inflect _ num _ T3 is 1, the welding spot is single-point spatter; when the inflect _ num _ T3 is 0, continuing to judge the characteristic variables R1 and R2, and if R1 is less than R2, the welding spot is normal; if R1 is equal to R2, the welding spot is in an electrodeless point state.
Optionally, a TPOT method is used to optimize a training process of the feature value set, and a classification model for classifying the welding spot states is obtained. The TPOT method is a Python automatic machine learning tool for optimizing a machine learning pipeline (pipeline) based on a genetic algorithm. And meanwhile, thousands of possible machine learning pipelines are intelligently explored, the optimal machine learning pipeline and the automatic super-parameter adjustment are obtained, a Python file of the optimal machine learning training pipeline code is automatically generated, the code file can be directly used for retraining a data source, and meanwhile, a prediction result and an evaluation index are automatically generated.
Step S105: and judging the welding spot state of the welding spot to be detected by utilizing the classification model.
Specifically, the judging process is as follows: acquiring process parameters of a welding spot to be detected in a welding process; constructing a resistance-time relation curve of the welding spot to be detected according to the technological parameters of the welding spot to be detected in the welding process; extracting a resistance minimum value, a resistance maximum value and the number of splash points from a resistance time relation curve of a welding spot to be detected to form a welding spot characteristic value set; inputting the characteristic value set of the welding spots to be detected into a classification model, and according to the characteristic values: and judging the state of the welding spot to which the welding spot to be detected belongs according to the resistance minimum value, the resistance maximum value and the number of the splashing points. The method realizes that the welding spot state can be quickly judged through the classification model when a new welding spot process parameter is input into the classification model by utilizing the pre-trained classification model.
Optionally, after the classification model is used to determine the state of the welding spot to be detected, the method further includes: and outputting the process parameters and the welding spot state of the welding spot under the condition that the judgment result of the welding spot state to be detected is abnormal. After the whole method completes the detection of mass welding spot states through automatic machine learning, the method provides data support for further analysis of welding spot performance, can find factors influencing the quality and performance of welding spots by combining sampling quality inspection analysis of welding spot real objects, and provides reliable data for providing welding spot process quality. And moreover, the welding spots with quality defects can be quickly and accurately found, main factors influencing quality abnormity can be timely searched and analyzed, and predictive maintenance such as quality inspection and process is realized, so that the aims of increasing efficiency, saving energy and reducing cost are fulfilled.
The present invention relates to a solder joint quality detection apparatus, and it should be noted that the judgment apparatus can be used for executing the solder joint quality detection method of the present invention.
Fig. 2 is a structural diagram of a welding spot quality detecting apparatus according to the present invention, and as shown in fig. 2, the welding spot quality detecting apparatus includes an obtaining module 10, a constructing module 20, an extracting module 30, a training module 40, and a judging module 50.
The acquisition module 10 is used for acquiring process parameters of a plurality of welding points in a welding process;
the building module 20 is used for building a resistance time relation curve of each welding point according to the process parameters;
the extraction module 30 is configured to extract a characteristic value of each welding point from the resistance-time relationship curve to obtain a welding point characteristic value set;
the training module 40 is used for training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar points; and
and the judging module 50 is configured to judge the welding spot state of the welding spot to be detected by using the classification model.
When the welding machine is used, the technological parameters of the target welding spot in the welding process are obtained through the obtaining module 10, wherein the technological parameters comprise the number, current, voltage, resistance and power of the welding spot and the corresponding welding time point, then a resistance time relation curve of the target welding spot is built through the building module 20 according to the technological parameters, and then the extraction module 30 extracts the characteristic value of the welding spot from the resistance time relation curve to obtain the characteristic value set of the welding spot; the training module 40 then uses the machine-learned classification model to finally determine the solder joint state of the target solder joint by using each solder joint feature value in the solder joint feature value set through the determination module 50, so as to achieve the determination of the solder joint state. Therefore, the processing of mass welding spot data is quickly and accurately finished, and the cost of welding spot quality detection is saved.
The present invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of: acquiring technological parameters of a plurality of welding points in the welding process; constructing a resistance time relation curve of each welding point according to the process parameters; extracting the characteristic value of each welding spot from the resistance-time relation curve to obtain a welding spot characteristic value set; training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar value points; and judging the welding spot state of the welding spot to be detected by utilizing the classification model.
Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing embodiments, or that equivalents may be substituted for elements thereof. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A welding spot quality detection method comprises the following steps:
acquiring technological parameters of a plurality of welding points in the welding process;
constructing a resistance time relation curve of each welding point according to the process parameters;
extracting the characteristic value of each welding spot from the resistance time relation curve to obtain a welding spot characteristic value set;
training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar points; and
and judging the welding spot state of the welding spot to be detected by utilizing the classification model.
2. The method of claim 1, wherein the extracting the characteristic value of each solder joint from the resistance-time relationship curve to obtain the solder joint characteristic value set comprises: extracting a resistance minimum value in a first time interval; extracting the maximum value of the resistance in the second time interval; extracting the resistance mutation points in the third time interval, and counting to obtain the number of the splash points; and forming a welding spot characteristic value set by the extracted resistance minimum value, the resistance maximum value and the number of the spattered points.
3. The solder joint quality detection method of claim 2, wherein the step of extracting the resistance discontinuity points in the third time interval and the step of counting the number of spattering points comprises the steps of: the resistance time relation curve is subjected to derivation to obtain a first derivative; acquiring a point which cannot be derived in a third time interval or a point at which the absolute value of the first derivative exceeds a set value; and counting the total number of the points which cannot be used for solving the derivative points and the points of which the absolute value of the first derivative exceeds the set value, so as to obtain the number of the splash points.
4. The welding spot quality detection method according to claim 2, wherein the first time interval is within 0-10ms from the welding start of each welding spot, the second time interval is within 20-30ms from the welding start of each welding spot, and the third time interval is within 30ms from the welding start to the welding end of each welding spot.
5. The method of claim 2, wherein the training of the weld feature value set of each weld and the weld state to which each weld belongs to obtain the classification model for classifying the weld states comprises: if the number of the spattering points exceeds one, outputting that the welding spot state is multipoint spattering; if the number of the spattering points is one, outputting that the welding spot state is single-point spattering; if the number of the splashing points is zero and the resistance minimum value is greater than or equal to the resistance maximum value, outputting a welding spot state as an electrodeless value point; and if the number of the splashing points is zero and the resistance minimum value is smaller than the resistance maximum value, outputting that the welding spot state is normal.
6. The method for detecting the quality of the welding spot according to claim 5, wherein the judging the welding spot state of the welding spot to be detected by using the classification model comprises: acquiring process parameters of a welding spot to be detected in a welding process; constructing a resistance-time relation curve of the welding spot to be detected according to the technological parameters of the welding spot to be detected in the welding process; extracting a resistance minimum value, a resistance maximum value and the number of splash points from a resistance time relation curve of a welding spot to be detected to form a welding spot characteristic value set; inputting the welding spot characteristic value set of the welding spot to be detected into the classification model, and according to the characteristic value: and judging the state of the welding spot to which the welding spot to be detected belongs according to the resistance minimum value, the resistance maximum value and the number of the splashing points.
7. The method for detecting the quality of the welding spot according to claim 2, wherein the specific way of training the welding spot feature value set of each welding spot and the welding spot state to which each welding spot belongs to obtain the classification model for classifying the welding spot state is as follows: and optimizing the training flow of the characteristic value set by using a TPOT (tire pressure test) method to obtain a classification model for classifying the states of the welding spots.
8. The method for detecting the quality of the welding spot, according to claim 1, wherein after acquiring the process parameters of the welding spots in the welding process, the method further comprises: and removing the technological parameters of the corresponding welding spots formed when the welding spot fails in the welding process and the welding gun is powered off.
9. The method according to claim 1, wherein after the classification model is used to determine the welding spot state of the welding spot to be detected, the method further comprises: and outputting the process parameters and the welding spot state of the welding spot under the condition that the judgment result of the welding spot state of the welding spot to be detected is abnormal.
10. A solder joint quality inspection apparatus for carrying out the method of claim 1, comprising:
the acquisition module is used for acquiring process parameters of a plurality of welding points in the welding process;
the construction module is used for constructing a resistance time relation curve of each welding point according to the process parameters;
the extraction module is used for extracting the characteristic value of each welding point from the resistance time relation curve to obtain a welding point characteristic value set;
the training module is used for training the welding spot characteristic value set of each welding spot and the welding spot state to which each welding spot belongs to obtain a classification model for classifying the welding spot states, wherein the welding spot states comprise multi-point splashing, single-point splashing, normal welding spots and non-polar points; and
and the judging module is used for judging the welding spot state of the welding spot to be detected by utilizing the classification model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110679149.7A CN113505657B (en) | 2021-06-18 | 2021-06-18 | Welding spot quality detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110679149.7A CN113505657B (en) | 2021-06-18 | 2021-06-18 | Welding spot quality detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113505657A true CN113505657A (en) | 2021-10-15 |
| CN113505657B CN113505657B (en) | 2022-05-03 |
Family
ID=78010528
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110679149.7A Active CN113505657B (en) | 2021-06-18 | 2021-06-18 | Welding spot quality detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113505657B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114083168A (en) * | 2021-11-17 | 2022-02-25 | 东风汽车集团股份有限公司 | Welding spatter point identification method and system |
| CN114330491A (en) * | 2021-11-24 | 2022-04-12 | 东风汽车集团股份有限公司 | Method for optimizing quality of welding spot through analysis guidance of resistance spot welding curve |
| CN114535768A (en) * | 2022-03-08 | 2022-05-27 | 东风汽车集团股份有限公司 | Resistance spot welding spatter identification method, device, equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101825580A (en) * | 2010-02-11 | 2010-09-08 | 上海交通大学 | Real-time detection method for resistance spot welding quality based on electrode displacement fluctuation |
| WO2012050108A1 (en) * | 2010-10-14 | 2012-04-19 | 住友金属工業株式会社 | Welding quality determination device |
| CN102458752A (en) * | 2009-04-07 | 2012-05-16 | 罗伯特·博世有限公司 | Method and control device for monitoring the quality of spot welds of a resistance welding gun comprising the outputting of a warning message |
| CN104457841A (en) * | 2014-11-04 | 2015-03-25 | 吉林大学 | Online monitoring method for resistance spot welding quality |
| CN110009011A (en) * | 2019-03-20 | 2019-07-12 | 华南理工大学 | A kind of solder joint recognition methods based on image procossing |
| WO2020048119A1 (en) * | 2018-09-04 | 2020-03-12 | Boe Technology Group Co., Ltd. | Method and apparatus for training a convolutional neural network to detect defects |
-
2021
- 2021-06-18 CN CN202110679149.7A patent/CN113505657B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102458752A (en) * | 2009-04-07 | 2012-05-16 | 罗伯特·博世有限公司 | Method and control device for monitoring the quality of spot welds of a resistance welding gun comprising the outputting of a warning message |
| CN101825580A (en) * | 2010-02-11 | 2010-09-08 | 上海交通大学 | Real-time detection method for resistance spot welding quality based on electrode displacement fluctuation |
| WO2012050108A1 (en) * | 2010-10-14 | 2012-04-19 | 住友金属工業株式会社 | Welding quality determination device |
| CN104457841A (en) * | 2014-11-04 | 2015-03-25 | 吉林大学 | Online monitoring method for resistance spot welding quality |
| WO2020048119A1 (en) * | 2018-09-04 | 2020-03-12 | Boe Technology Group Co., Ltd. | Method and apparatus for training a convolutional neural network to detect defects |
| CN110009011A (en) * | 2019-03-20 | 2019-07-12 | 华南理工大学 | A kind of solder joint recognition methods based on image procossing |
Non-Patent Citations (1)
| Title |
|---|
| 史建涛等: "基于电极位移信号的电阻点焊质量预测模型", 《电焊机》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114083168A (en) * | 2021-11-17 | 2022-02-25 | 东风汽车集团股份有限公司 | Welding spatter point identification method and system |
| CN114083168B (en) * | 2021-11-17 | 2022-11-29 | 东风汽车集团股份有限公司 | Welding spatter point identification method and system |
| CN114330491A (en) * | 2021-11-24 | 2022-04-12 | 东风汽车集团股份有限公司 | Method for optimizing quality of welding spot through analysis guidance of resistance spot welding curve |
| CN114330491B (en) * | 2021-11-24 | 2024-08-02 | 东风汽车集团股份有限公司 | Method for optimizing welding spot quality by analyzing and guiding resistance spot welding curve |
| CN114535768A (en) * | 2022-03-08 | 2022-05-27 | 东风汽车集团股份有限公司 | Resistance spot welding spatter identification method, device, equipment and storage medium |
| CN114535768B (en) * | 2022-03-08 | 2023-05-26 | 东风汽车集团股份有限公司 | Resistance spot welding splash identification method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113505657B (en) | 2022-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113505657B (en) | Welding spot quality detection method and device | |
| CN111047225B (en) | SMT surface mounting component welding spot quality evaluation method based on edge side model processing | |
| CN111300144B (en) | Automatic detection method for tool wear state based on image processing | |
| JP5632009B2 (en) | Method and apparatus for monitoring welding quality | |
| US6441342B1 (en) | Monitor for electric arc welder | |
| CN112692646B (en) | Intelligent assessment method and device for tool wear state | |
| CN101364106A (en) | Welding quality control system and method for resistance welding | |
| CN114330491B (en) | Method for optimizing welding spot quality by analyzing and guiding resistance spot welding curve | |
| CN101566546A (en) | Online weld inspection and repair method for resistance welding and weld-bonding | |
| CN114871620B (en) | Thick plate welding quality monitoring and control system and welding method thereof | |
| Casoetto et al. | Multisensor process performance assessment through use of autoregressive modeling and feature maps | |
| CN107335908B (en) | Automated stochastic method for feature discovery and its use in iterative processes | |
| CN111331211B (en) | On-line penetration detection method and system for electric spark small hole machining | |
| Lee et al. | Development of real-time diagnosis framework for angular misalignment of robot spot-welding system based on machine learning | |
| CN116618878A (en) | Pre-welding process parameter determination method, welding quality online prediction method, device and storage medium | |
| JP2004529416A (en) | How to regulate and control technical processes | |
| CN117961646B (en) | Method and system for predicting residual service life of numerical control blade for heavy load machining | |
| CN112025064A (en) | Resistance spot welding short-time amplitude modulation control method based on spatter statistical feedback | |
| CN113553659B (en) | System and method for fitting standard curve band of welding spot | |
| CN111914473B (en) | Welding parameter determining method and device for resistance spot welding, electronic equipment and storage medium | |
| Russell et al. | Comparison and explanation of data-driven modeling for weld quality prediction in resistance spot welding | |
| CN110154027B (en) | Robot intelligent welding method based on cloud computing | |
| CN115255566B (en) | Welding deviation real-time intelligent detection method based on high-quality time domain characteristics | |
| CN113487149A (en) | Welding spot abnormity identification system and method based on Catboost K-fold cross verification | |
| Lin et al. | Welding quality analysis and prediction based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |