Double-cavity laser welding fall-off risk warning method

Abstract

The present invention discloses a dual-cavity mode laser welding fall-off risk warning method, which relates to the field of welding management technology. The method comprises: performing a correlation analysis of the first welding preset influencing factor on welding fall-off, and obtaining a welding fall-off risk factor that does not include the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency; performing high-frequency assignment on the welding fall-off risk factor to obtain a standard value. Based on the standard value, weights are assigned to the laser wavelength, output interval duration, and output frequency, and a hidden Markov chain prediction model is configured as a welding fall-off risk prediction function. Welding is performed and monitoring values ​​are obtained, and these monitoring values ​​are mapped to welding fall-off probabilities through a welding fall-off risk prediction function. When the welding fall-off probability reaches or exceeds a set threshold, an audible and visual alarm is used to issue a warning. Thereby, a feasible and reliable warning path is provided, and the technical effect of improving the accuracy of the warning is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of welding management, and in particular to a double-cavity mold laser welding shedding risk early warning method. Background Art

[0002] Due to its unique structure and welding method, dual-cavity laser welding can improve welding quality and efficiency. Correspondingly, this welding technology has extremely high requirements for the control of welding parameters. The welding process involves many factors such as base material, welding powder, equipment, and weld design.

[0003] The current welding control mainly relies on experience and traditional detection technology, which makes it difficult to respond to the complex and changing welding parameters in the dual-cavity laser welding process in real time. There are technical problems such as difficulty in early warning, insufficient reliability and accuracy. Summary of the invention

[0004] The present invention provides a dual-cavity mold laser welding detachment risk warning method to solve the technical problems of high warning difficulty, insufficient reliability and accuracy in the prior art, and achieve the technical effect of providing a feasible and reliable warning path and improving the accuracy of the warning.

[0005] The double-cavity mode laser welding falling risk early warning method provided by the present invention comprises:

[0006] A correlation analysis is performed on the first welding preset influencing factor to obtain a welding fall-off risk factor, wherein the first welding preset influencing factor does not include the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency.

[0007] The welding fall-off risk factor is assigned a high frequency value to obtain a standard value of the welding fall-off risk factor.

[0008] Based on the standard value of the welding fall-off risk factor, weights are allocated to the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency to obtain a weight allocation result.

[0009] Based on the standard value of the welding fall-off risk factor and the weight distribution result, a hidden Markov chain prediction model of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, the dual-cavity mode laser output frequency and the welding fall-off probability is configured and set as a welding fall-off risk prediction function.

[0010] After the dual-cavity mode laser is initialized with the standard value of the welding fall-off risk factor, welding is started to obtain the first cavity mode laser wavelength monitoring value, the second cavity mode laser wavelength monitoring value, the dual-cavity mode output interval duration monitoring value and the dual-cavity mode laser output frequency monitoring value.

[0011] The welding fall-off probability is generated by mapping the first cavity mode laser wavelength monitoring value, the second cavity mode laser wavelength monitoring value, the dual-cavity mode output interval duration monitoring value and the dual-cavity mode laser output frequency monitoring value through the welding fall-off risk prediction function.

[0012] When the welding fall-off probability is greater than or equal to the welding fall-off probability threshold, an early warning is performed through the sound and light alarm of the dual-cavity mode laser.

[0013] The present invention discloses a dual-cavity mode laser welding fall-off risk warning method, comprising: performing a correlation analysis on a first welding preset influencing factor to identify a welding fall-off risk factor, but excluding the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency; performing high-frequency assignment on the welding fall-off risk factor to obtain its standard value; based on the standard value, performing weight assignment on the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency to obtain a weight assignment result; based on the standard value of the welding fall-off risk factor and the weight assignment result , establish a hidden Markov chain prediction model, associate the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration and the dual cavity mode laser output frequency with the welding fall-off probability, and form a welding fall-off risk prediction function; start welding after the dual cavity mode laser is initialized, and obtain monitoring values: the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration and the dual cavity mode laser output frequency; map the monitoring values ​​through the welding fall-off risk prediction function to generate the welding fall-off probability; if the probability is greater than or equal to the set welding fall-off probability threshold, then issue an early warning through the sound and light alarm of the dual cavity mode laser. The dual cavity mode laser welding fall-off risk early warning method disclosed in the present invention solves the technical problems of high difficulty in early warning, insufficient reliability and accuracy, and achieves the technical effect of providing a feasible and reliable early warning path and improving the accuracy of early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic diagram of the process of the double-cavity mold laser welding falling risk early warning method of the present invention;

[0015] Figure 2 It is a schematic diagram of the process of obtaining the welding fall-off risk factor in the dual-cavity mode laser welding fall-off risk warning method of the present invention. DETAILED DESCRIPTION

[0016] The technical solution provided in the embodiments of the present invention is to solve the technical problems of the existing technology in that the early warning is difficult and the reliability and accuracy are insufficient. The overall idea adopted is as follows:

[0017] Firstly, the correlation analysis of welding fall-off of the first welding preset influencing factor is carried out to determine the welding fall-off risk factor, and the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency are excluded; then, the welding fall-off risk factor is assigned with high frequency to obtain its standard value; then, based on the standard value of the welding fall-off risk factor, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency are weighted to obtain the weighted distribution result; then, with the standard value of the welding fall-off risk factor and the weighted distribution result as restrictions, the hidden Markov chain prediction model is configured and set as the welding fall-off risk prediction function; then, the dual-cavity mode laser is initialized before welding to obtain the monitoring values ​​of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency; finally, the monitoring values ​​are mapped through the welding fall-off risk prediction function to generate the welding fall-off probability. When the probability is greater than or equal to the set threshold, an early warning is issued through the sound and light alarm.

[0018] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them. Example

[0019] Figure 1 The figure is a flow chart of the double-cavity mold laser welding fall-off risk warning method of the present invention, wherein the method comprises:

[0020] A correlation analysis is performed on the first welding preset influencing factor to obtain a welding fall-off risk factor, wherein the first welding preset influencing factor does not include the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency.

[0021] Specifically, the base material, welding powder, equipment, weld design, welding parameters, etc. in laser welding will all have an impact on the welding situation. In particular, in dual-cavity laser welding, the welding control parameters of the dual-cavity mode are more complex, and the control refinement requirements are higher. They are mostly adjusted in real time by the equipment control components according to the welding situation, and there is a large uncertainty, which makes the influencing factors including the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency present a very complex nonlinear mapping relationship with the welding detachment risk, which is difficult to predict. Therefore, when conducting a correlation analysis of welding detachment, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency are excluded.

[0022] Optionally, the first welding preset influencing factors include the chemical composition and mechanical properties of the base material and welding wire, the type and flow rate of the shielding gas used, the temperature and humidity conditions of the connection environment, the weld design, etc.

[0023] In some embodiments, Figure 2 As shown, the correlation analysis of welding fall-off is performed on the first welding preset influencing factor to obtain the welding fall-off risk factor, including:

[0024] In response to a welding request from a user end, basic welding information is obtained, wherein the basic welding information includes welding parent materials and welding condition type numbers; based on the welding parent materials and the welding condition type numbers, a one-to-one corresponding first welding preset influencing factor record data set and a first welding fall-off identification data set are collected, wherein the first welding fall-off identification data set includes 0 or 1, 1 represents fall-off within a set service time, and 0 represents no fall-off within the set service time; the first welding preset influencing factor record data set is clustered in a consistent manner to generate multiple groups of first welding preset influencing factor record data, wherein consistent clustering refers to clustering into one category the deviations of each attribute welding preset influencing factor that are less than or equal to a welding preset influencing factor deviation threshold; the first welding fall-off identification data set is grouped according to the multiple groups of first welding preset influencing factor record data to obtain multiple groups of first welding fall-off identification data; correlation analysis is performed according to the multiple groups of first welding preset influencing factor record data and the multiple groups of first welding fall-off identification data sets to obtain the welding fall-off risk factor.

[0025] Specifically, firstly, the welding request from the user is received, and the relevant basic welding information is obtained, including the type of welding base material (such as steel, aluminum, etc.) and the welding condition type number. Among them, the welding condition type number is used to call the corresponding welding position information (such as horizontal, vertical, etc.) and path information.

[0026] Specifically, based on the obtained welding base material and welding condition type number, the corresponding first welding preset influencing factor record data set and the first welding detachment identification data set are collected, wherein the first welding preset influencing factor record data set is a structured data set that records all influencing factors related to the welding process (such as welding current, welding speed, etc.). The first welding detachment identification data set is a corresponding data set that identifies the welding detachment situation. Exemplarily, the first welding detachment identification data set includes multiple Boolean values, such as 0 and 1, wherein 1 indicates that welding detachment occurred within the set service time, and 0 indicates that detachment did not occur.

[0027] Specifically, by classifying the elements whose deviation of each welding preset influencing factor is within the set welding preset influencing factor deviation threshold into one category, the first welding preset influencing factor record data set is clustered consistently, and similar influencing factors are classified through the above steps, so that complex discrete data can be clustered into structured data with strong representativeness, which helps to reduce the complexity and redundancy of the data.

[0028] Specifically, according to the generated multiple sets of first welding preset influencing factor record data, the corresponding first welding peeling identification data set is called for grouping to form multiple sets of first welding peeling identification data, wherein each set of first welding peeling identification data corresponds to a set of first welding preset influencing factor record data.

[0029] Furthermore, based on statistical methods (such as correlation analysis, regression analysis, etc.), correlation analysis is performed between multiple groups of first welding preset influencing factor record data and multiple groups of first welding fall-off identification data sets, and welding fall-off risk factors are identified and extracted based on the correlation analysis results. Among them, the correlation analysis results are used to quantitatively indicate which welding preset influencing factors have a significant impact on welding fall-off, thereby providing a usable warning dimension direction for subsequent warnings. Through the above steps, the factors that may cause fall-off during the welding process can be systematically analyzed, thereby achieving more efficient risk management and control.

[0030] In some implementations, performing correlation analysis on the plurality of sets of first welding preset influencing factor record data and the plurality of sets of first welding fall-off identification data sets to obtain the welding fall-off risk factor includes:

[0031] A first welding preset influencing factor record data is randomly extracted from each of the multiple groups of first welding preset influencing factor record data to obtain multiple first welding preset influencing factor record data; the multiple groups of first welding fall-off identification data sets are traversed to add identification values ​​to obtain multiple first identification value addition results; the multiple first identification value addition results are used as a reference sequence, and the multiple first welding preset influencing factor record data are used as multiple comparison sequences to perform grey correlation analysis to obtain multiple grey correlation degrees; according to the multiple grey correlation degrees, the welding fall-off risk factors that are greater than or equal to a grey correlation degree threshold are sorted from the first welding preset influencing factor.

[0032] Specifically, first, a plurality of first welding preset influencing factor record data are randomly selected from a plurality of first welding preset influencing factor record data, wherein the plurality of first welding preset influencing factor record data are sample data of a class of preset influencing factors. Then, according to the plurality of first welding preset influencing factor record data obtained, a plurality of first welding fall-off identification data sets are traversed, and the identification values ​​of each group are summed up to obtain the sum of a plurality of first identification values. This result reflects the fall-off situation under different welding conditions.

[0033] Specifically, the sum of multiple first identification values ​​is used as the reference sequence, and multiple first welding preset influencing factor record data is used as the comparison sequence to perform grey correlation analysis. Exemplarily, it includes normalizing the reference sequence and the comparison sequence; calculating the difference sequence between the reference sequence and the comparison sequence; calculating the absolute value of the difference sequence; calculating the correlation coefficient: calculating the correlation coefficient at each moment according to the absolute difference; averaging the correlation coefficients to obtain the grey correlation. The obtained grey correlation evaluates the strength of the relationship between each type of welding preset influencing factor and welding detachment.

[0034] Furthermore, based on the multiple grey correlation degrees obtained, the welding preset influencing factors whose grey correlation degrees are greater than or equal to the set grey correlation degree threshold are screened out. These screened out factors are welding fall-off risk factors, which can be considered to have a strong correlation with welding fall-off.

[0035] The welding fall-off risk factor is assigned a high frequency value to obtain a standard value of the welding fall-off risk factor.

[0036] Specifically, the standard value of the welding peeling risk factor refers to the typical value of the welding peeling risk factor, that is, the data point or data interval where each welding peeling risk factor appears frequently.

[0037] In some embodiments, high-frequency assignment is performed on the welding fall-off risk factor to obtain a standard value of the welding fall-off risk factor, including:

[0038] In response to a welding request from a user end, basic welding information is obtained, wherein the basic welding information includes welding base materials and welding condition type numbers; taking the welding base materials and the welding condition type numbers as constraints and the welding fall-off risk factor as a retrieval target, a welding sample set whose service time of a weld is greater than or equal to a set service time is collected; a first attribute welding fall-off risk factor characteristic value set of the welding sample set is obtained, a central trend analysis is performed on the first attribute welding fall-off risk factor characteristic value set, and a concentrated distribution interval of the first attribute welding fall-off risk factor is obtained, which is set as a first attribute welding fall-off risk factor standard value; and the first attribute welding fall-off risk factor standard value is added to the welding fall-off risk factor standard value.

[0039] Specifically, the welding request from the user is parsed, the welding base material and welding condition type number are obtained as index constraints, the welding record database of the target scene or the target scene is traversed, and the welding sample set whose service time of the weld is greater than or equal to the set service time is extracted, wherein the welding sample set includes the historical record values ​​of multiple types of welding fall-off risk factors. The service time is set to the expected durability of the weld in the target scene or target product.

[0040] Specifically, multiple welding fall-off risk factor historical record values ​​corresponding to any welding fall-off risk factor are extracted from the welding sample set to obtain the first attribute welding fall-off risk factor characteristic value set. Then, the first attribute welding fall-off risk factor characteristic value set is statistically analyzed to calculate the main central tendency indicators, such as mean, median and mode. The above-mentioned main central tendency indicators can reflect the central position of the characteristic value, that is, the high-frequency position of the characteristic value. Then, according to the calculated central tendency indicator, a reasonable range is set, which is the concentrated distribution interval of the first attribute welding fall-off risk factor, indicating the expected range of change of the risk factor under normal welding conditions. Exemplarily, the standard value interval of the first attribute welding fall-off risk factor is defined by mean ± k × standard deviation (k is a constant, usually 1 or 2).

[0041] Optionally, display the distribution of feature values ​​through visualization tools such as histograms or box plots to help identify their concentration and dispersion.

[0042] Furthermore, the standard value of the first attribute welding peeling risk factor is added to the standard value of the welding peeling risk factor, and the above extraction steps are repeated to obtain the welding peeling risk factor characteristic value set for central trend analysis, and generate multiple welding peeling risk factor standard values ​​to add to the welding peeling risk factor standard value.

[0043] Based on the standard value of the welding fall-off risk factor, weights are allocated to the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency to obtain a weight allocation result.

[0044] In a feasible implementation, based on the standard value of the welding fall-off risk factor, weight allocation is performed on the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency to obtain a weight allocation result, including:

[0045] The standard value of the welding peeling risk factor, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency are sent to several evaluation nodes that cannot communicate with each other to obtain several feedback information; the several feedback information include several first cavity mode laser wavelength importance scores, several second cavity mode laser wavelength importance scores, several dual-cavity mode output interval duration importance scores and several dual-cavity mode laser output frequency importance scores; according to the several first cavity mode laser wavelength importance scores, the several second cavity mode laser wavelength importance scores, the several dual-cavity mode output interval duration importance scores and the several dual-cavity mode laser output frequency importance scores, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency are weighted to obtain the weighted allocation result.

[0046] Specifically, the standard value of the welding fall-off risk factor, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration, and the dual cavity mode laser output frequency are sent to multiple evaluation nodes that cannot communicate with each other. Among them, the standard value of the welding fall-off risk factor can be considered as an evaluation constraint of multiple evaluation nodes, which is used to initialize the evaluation. Multiple evaluation nodes are different evaluation algorithms or rules, expert systems or independent computing units, aiming to obtain independent feedback.

[0047] Specifically, each evaluation node independently evaluates the received information and generates a number of feedback information, including: the importance score of the first cavity mode laser wavelength, which is used to characterize the degree of influence of the first cavity mode laser wavelength on the risk of welding detachment. The importance score of the second cavity mode laser wavelength, which is used to characterize the influence of the second cavity mode laser wavelength. The importance score of the dual cavity mode output interval duration, which is used to characterize the influence of the dual cavity mode output interval duration on the welding quality. The importance score of the dual cavity mode laser output frequency, which is used to characterize the influence of the dual cavity mode laser output frequency on the risk of welding detachment.

[0048] Furthermore, the information fed back by all evaluation nodes is integrated through normalization processing, and the weight of each laser parameter is determined according to the integrated importance score.

[0049] In some implementations, weighting the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration, and the dual cavity mode laser output frequency is performed according to the plurality of first cavity mode laser wavelength importance scores, the plurality of second cavity mode laser wavelength importance scores, the plurality of dual cavity mode output interval duration importance scores, and the plurality of dual cavity mode laser output frequency importance scores to obtain the weighting result, including:

[0050] Add the several first cavity mode laser wavelength importance scores, the several second cavity mode laser wavelength importance scores, the several dual-cavity mode output interval duration importance scores and the several dual-cavity mode laser output frequency importance scores to obtain the total score; traverse the several first cavity mode laser wavelength importance scores, the several second cavity mode laser wavelength importance scores, the several dual-cavity mode output interval duration importance scores and the several dual-cavity mode laser output frequency importance scores and add them respectively to obtain the first cavity mode laser wavelength importance summed score, the second cavity mode laser wavelength importance summed score, the dual-cavity mode output interval duration importance summed score and the dual-cavity mode laser output frequency importance summed score; traverse the first cavity mode laser wavelength importance summed score, the second cavity mode laser wavelength importance summed score, the dual-cavity mode output interval duration importance summed score and the dual-cavity mode laser output frequency importance summed score, compare with the total score to obtain the weight distribution result.

[0051] Specifically, firstly, all the first cavity mode laser wavelength importance scores, second cavity mode laser wavelength importance scores, dual cavity mode output interval duration importance scores and dual cavity mode laser output frequency importance scores are added to obtain the total score; then, the first cavity mode laser wavelength importance scores, second cavity mode laser wavelength importance scores, dual cavity mode output interval duration importance scores and dual cavity mode laser output frequency importance scores are classified and added to obtain the total classification score of each type of control parameter, that is, the first cavity mode laser wavelength importance sum score, the second cavity mode laser wavelength importance sum score, the dual cavity mode output interval duration importance sum score and the dual cavity mode laser output frequency importance sum score. Finally, through normalization processing, the sum of each classification score is divided by the sum of the scores to obtain the weight of each type of control parameter, and the sum of the weights is ensured to be 1.

[0052] Based on the standard value of the welding fall-off risk factor and the weight distribution result, a hidden Markov chain prediction model of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, the dual-cavity mode laser output frequency and the welding fall-off probability is configured and set as a welding fall-off risk prediction function.

[0053] In some embodiments, based on the standard value of the welding fall-off risk factor and the weight distribution result, a hidden Markov chain prediction model of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency and the welding fall-off probability is configured and set as a welding fall-off risk prediction function, including:

[0054] Based on the standard value of the welding fall-off risk factor and the weight distribution result, a state transition probability matrix of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency, and the welding fall-off probability is constructed; supervised training is performed on the hidden Markov chain topology structure according to the state transition probability matrix to generate the welding fall-off risk prediction function.

[0055] Specifically, firstly, according to the standard value and weight distribution results of the welding fall-off risk factor, the four variables of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration and the dual cavity mode laser output frequency are incorporated into the model. The different values ​​of each variable are combined with the corresponding welding fall-off probability to form a state transition probability matrix. And the probability of transitioning from one state to another under specific conditions is identified.

[0056] Specifically, the known welding fall-off marks (0 and 1) in the collected welding log record data are used to guide the model to learn the transition probability between states for supervised training. During the training process, the state transition probability is optimized multiple times to ensure that the model accurately reflects the impact of laser parameters on the fall-off probability during welding. The welding fall-off risk prediction function obtained after training will be able to map the given laser wavelength, output interval duration, and output frequency to the corresponding welding fall-off probability.

[0057] Through the above-mentioned welding fall-off risk prediction function, an accurate mathematical model is provided for the prediction of welding fall-off risk, which helps to enhance the prediction ability of welding fall-off risk and thus optimize the welding process.

[0058] In some implementations, based on the standard value of the welding fall-off risk factor and the weight distribution result, a state transition probability matrix of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency, and the welding fall-off probability is constructed, including:

[0059] With the welding fall-off risk factor standard value as a limit, a welding log record data set is collected, wherein the welding log record data set includes a first cavity mode laser wavelength record data set, a second cavity mode laser wavelength record data set, a dual-cavity mode output interval duration record data set, a dual-cavity mode laser output frequency record data set and a second welding fall-off identification data set, wherein the second welding fall-off identification data set includes 0 or 1, 1 indicates that the fall-off occurs within the set service time, and 0 indicates that the fall-off does not occur within the set service time; the first cavity mode laser wavelength record data set, the second cavity mode laser wavelength record data set, the dual-cavity mode output interval duration record data set, the dual-cavity mode laser output frequency record data set and the second welding fall-off identification data set. The recording data set and the dual-cavity mode laser output frequency recording data set are clustered in consistency to obtain multiple groups of welding log recording data; the second welding fall-off identification data are grouped according to the multiple groups of welding log recording data to obtain multiple groups of second welding fall-off identification data; the multiple groups of second welding fall-off identification data are traversed to count the proportion of the number of 0-value identifications to obtain multiple welding fall-off probabilities; the multiple groups of welding log recording data are weighted according to the weight distribution result to construct multiple groups of welding control matrices; the state transition probability matrix is ​​constructed according to the multiple groups of welding control matrices and the multiple welding fall-off probabilities.

[0060] Specifically, first, the standard value of the welding fall-off risk factor is used as a retrieval constraint to match and obtain the welding log record data set; then, based on the same method principle as the consistency clustering of the first welding preset influencing factor record data set, the welding log record data set is clustered consistently to reduce the complexity and redundant data of the data set and improve the data quality; then, according to the grouping results obtained by the consistency clustering, multiple groups of second welding fall-off identification data are obtained, and the multiple groups of second welding fall-off identification data are data subsets of multiple groups of welding log record data. Finally, the proportion of the number of 0-value identifications in the multiple groups of second welding fall-off identification data is counted, and the output is multiple welding fall-off probabilities.

[0061] Specifically, the multiple welding control matrices are matrices that take into account the influence of different welding control parameter weights, through which the contribution of different welding control parameters in different welding log record data to the welding fall-off probability can be defined. Exemplarily, the multiple welding control matrices are one-dimensional matrices that reflect the influence of different welding control parameters.

[0062] Furthermore, multiple sets of welding control matrices are combined with multiple welding dropout probabilities to construct a state transition probability matrix. The state transition probability matrix characterizes the possibility of welding dropout under specific welding conditions and forms the basis of the hidden Markov model. Through the above steps, the established state transition probability matrix is ​​used to provide data support for the prediction and control of welding dropout.

[0063] After the dual-cavity mode laser is initialized with the standard value of the welding fall-off risk factor, welding is started to obtain the first cavity mode laser wavelength monitoring value, the second cavity mode laser wavelength monitoring value, the dual-cavity mode output interval duration monitoring value and the dual-cavity mode laser output frequency monitoring value.

[0064] Specifically, taking the standard value of the welding fall-off risk factor as the parameter selection space, the initial welding environment of the dual-cavity mode laser is configured, including setting the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency as initial values, to ensure that the laser is in the best working state and the accuracy of subsequent warnings.

[0065] Specifically, after initialization is completed, the dual-cavity mode laser is activated for welding, and the changes in the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency are synchronously collected. Preferably, the above monitoring values ​​are obtained based on the same collection accuracy to effectively obtain the key monitoring values ​​in the welding process and reduce monitoring overhead.

[0066] The welding fall-off probability is generated by mapping the first cavity mode laser wavelength monitoring value, the second cavity mode laser wavelength monitoring value, the dual-cavity mode output interval duration monitoring value and the dual-cavity mode laser output frequency monitoring value through the welding fall-off risk prediction function.

[0067] Furthermore, the first cavity mode laser wavelength monitoring value, the second cavity mode laser wavelength monitoring value, the dual-cavity mode output interval duration monitoring value and the dual-cavity mode laser output frequency monitoring value are input into the welding fall-off risk prediction function to obtain the corresponding mapping value, and the output is the welding fall-off probability.

[0068] When the welding fall-off probability is greater than or equal to the welding fall-off probability threshold, an early warning is performed through the sound and light alarm of the dual-cavity mode laser.

[0069] Furthermore, the calculated welding fall-off probability is compared with a preset welding fall-off probability threshold. If the welding fall-off probability is greater than or equal to the threshold, it indicates that there is a high risk of fall-off in the current welding process. The sound and light alarm of the dual-cavity mode laser is activated to sound an alarm to remind the operator, which helps to take timely measures to avoid further deterioration of welding quality problems and avoid waste of resources caused by ineffective operations.

[0070] In summary, the dual-cavity laser welding fall-off risk warning method provided by the present invention has the following technical effects:

[0071] The correlation analysis of the first welding preset influencing factors is performed to identify the welding fall-off risk factor, but does not include the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency; the welding fall-off risk factor is assigned a high frequency value to obtain its standard value; based on the standard value, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval time and the dual-cavity mode laser output frequency are weighted to obtain the weight distribution result; based on the standard value of the welding fall-off risk factor and the weight distribution result, a hidden Markov chain prediction is established. The model associates the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration and the dual cavity mode laser output frequency with the welding shedding probability to form a welding shedding risk prediction function; after the dual cavity mode laser is initialized, welding is started to obtain monitoring values: the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration and the dual cavity mode laser output frequency; the monitoring values ​​are mapped through the welding shedding risk prediction function to generate the welding shedding probability; if the probability is greater than or equal to the set welding shedding probability threshold, an early warning is issued through the sound and light alarm of the dual cavity mode laser. Thus, a feasible and reliable early warning path is provided to improve the accuracy of early warning.

[0072] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A double-cavity laser welding falling risk warning method, characterized in that: include: Performing a correlation analysis on the welding fall-off of the first welding preset influencing factor to obtain a welding fall-off risk factor, wherein the first welding preset influencing factor does not include the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency; Performing statistical analysis and assigning values ​​to the welding fall-off risk factor to obtain a standard value of the welding fall-off risk factor; Based on the standard value of the welding fall-off risk factor, weight allocation is performed on the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency to obtain a weight allocation result; Based on the standard value of the welding fall-off risk factor and the weight distribution result, a hidden Markov chain prediction model of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, the dual-cavity mode laser output frequency and the welding fall-off probability is configured and set as a welding fall-off risk prediction function; After the dual-cavity mode laser is initialized with the standard value of the welding fall-off risk factor, welding is started to obtain a first cavity mode laser wavelength monitoring value, a second cavity mode laser wavelength monitoring value, a dual-cavity mode output interval duration monitoring value, and a dual-cavity mode laser output frequency monitoring value; The first cavity mode laser wavelength monitoring value, the second cavity mode laser wavelength monitoring value, the dual-cavity mode output interval duration monitoring value and the dual-cavity mode laser output frequency monitoring value are mapped by the welding fall-off risk prediction function to generate a welding fall-off probability; When the welding fall-off probability is greater than or equal to the welding fall-off probability threshold, an early warning is performed through the sound and light alarm of the dual-cavity mode laser.

2. The double-cavity laser welding falling risk warning method according to claim 1, characterized in that: The correlation analysis of welding fall-off of the first welding preset influencing factors is carried out to obtain the welding fall-off risk factors, including: In response to a welding request from a user terminal, basic welding information is obtained, wherein the basic welding information includes a welding base material and a welding condition type number; Based on the welding base material and the welding condition type number, a first welding preset influencing factor record data set and a first welding fall-off identification data set corresponding to each other are collected, wherein the first welding fall-off identification data includes 0 or 1, 1 indicates that the welding falls off within the set service time, and 0 indicates that the welding falls off within the set service time; Performing consistency clustering on the first welding preset influencing factor record data set to generate multiple groups of first welding preset influencing factor record data, wherein the consistency clustering refers to clustering the welding preset influencing factor deviations of each attribute that are less than or equal to the welding preset influencing factor deviation threshold into one category; The first welding peeling identification data set is grouped according to the multiple groups of first welding preset influencing factor record data to obtain multiple groups of first welding peeling identification data; The welding fall-off risk factor is obtained by performing a correlation analysis based on the multiple groups of first welding preset influencing factor record data and the multiple groups of first welding fall-off identification data sets.

3. The double-cavity laser welding falling risk warning method according to claim 2, characterized in that: The welding fall-off risk factor is obtained by performing correlation analysis based on the plurality of sets of first welding preset influencing factor record data and the plurality of sets of first welding fall-off identification data sets, including: randomly extracting one first welding preset influencing factor record data from each of the plurality of groups of first welding preset influencing factor record data to obtain a plurality of first welding preset influencing factor record data; Traversing the plurality of groups of first welding peeling identification data sets and respectively adding identification values ​​to obtain a plurality of first identification value addition results; Taking the sum of the plurality of first identification values ​​as a reference sequence and taking the plurality of first welding preset influencing factor record data as a plurality of comparison sequences, a grey relational analysis is performed to obtain a plurality of grey relational degrees; According to the multiple grey correlation degrees, the welding peeling risk factors greater than or equal to a grey correlation degree threshold are sorted from the first welding preset influencing factors.

4. The double-cavity laser welding fall-off risk warning method according to claim 1, characterized in that: Performing statistical analysis and assigning values ​​to the welding fall-off risk factor to obtain a standard value of the welding fall-off risk factor includes: In response to a welding request from a user terminal, basic welding information is obtained, wherein the basic welding information includes a welding base material and a welding condition type number; Taking the welding base material and the welding condition type number as constraints and the welding fall-off risk factor as a search target, a welding sample set whose service time of the welding spot is greater than or equal to the set service time is collected; Obtain a first attribute welding fall-off risk factor characteristic value set of the welding sample set, perform a central tendency analysis on the first attribute welding fall-off risk factor characteristic value set, obtain a first attribute welding fall-off risk factor concentrated distribution interval, and set it as a first attribute welding fall-off risk factor standard value; The first attribute welding peeling risk factor standard value is added to the welding peeling risk factor standard value.

5. The double-cavity laser welding falling risk warning method according to claim 1, characterized in that: Based on the standard value of the welding fall-off risk factor, weight allocation is performed on the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency to obtain a weight allocation result, including: Sending the welding peeling risk factor standard value, the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, and the dual-cavity mode laser output frequency to a plurality of evaluation nodes that cannot communicate with each other to obtain a plurality of feedback information; The plurality of feedback information includes a plurality of first cavity mode laser wavelength importance scores, a plurality of second cavity mode laser wavelength importance scores, a plurality of dual cavity mode output interval duration importance scores and a plurality of dual cavity mode laser output frequency importance scores; According to the importance scores of the several first cavity mode laser wavelengths, the importance scores of the several second cavity mode laser wavelengths, the importance scores of the several dual-cavity mode output interval durations and the importance scores of the several dual-cavity mode laser output frequencies, weights are allocated to the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency to obtain the weight allocation result.

6. The double-cavity laser welding fall-off risk warning method according to claim 5, characterized in that: The first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual cavity mode output interval duration and the dual cavity mode laser output frequency are weighted according to the plurality of first cavity mode laser wavelength importance scores, the plurality of second cavity mode laser wavelength importance scores, the plurality of dual cavity mode output interval duration importance scores and the plurality of dual cavity mode laser output frequency importance scores to obtain the weighted allocation result, including: Adding the importance scores of the plurality of first cavity mode laser wavelengths, the importance scores of the plurality of second cavity mode laser wavelengths, the importance scores of the plurality of dual-cavity mode output interval durations, and the importance scores of the plurality of dual-cavity mode laser output frequencies to obtain a total score; Traversing the plurality of first cavity mode laser wavelength importance scores, the plurality of second cavity mode laser wavelength importance scores, the plurality of dual-cavity mode output interval duration importance scores and the plurality of dual-cavity mode laser output frequency importance scores, and respectively adding them up, to obtain a first cavity mode laser wavelength importance summed score, a second cavity mode laser wavelength importance summed score, a dual-cavity mode output interval duration importance summed score and a dual-cavity mode laser output frequency importance summed score; The weight distribution result is obtained by traversing the first cavity mode laser wavelength importance sum score, the second cavity mode laser wavelength importance sum score, the dual-cavity mode output interval duration importance sum score and the dual-cavity mode laser output frequency importance sum score, and comparing them with the total score.

7. The double-cavity laser welding fall-off risk warning method according to claim 1, characterized in that: Based on the standard value of the welding fall-off risk factor and the weight distribution result, a hidden Markov chain prediction model of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, the dual-cavity mode laser output frequency and the welding fall-off probability is configured and set as a welding fall-off risk prediction function, including: Based on the standard value of the welding fall-off risk factor and the weight distribution result, a state transition probability matrix of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration, the dual-cavity mode laser output frequency, and the welding fall-off probability is constructed; The hidden Markov chain topology structure is supervised and trained according to the state transition probability matrix to generate the welding fall-off risk prediction function.

8. The double-cavity laser welding fall-off risk warning method according to claim 7, characterized in that: Based on the standard value of the welding fall-off risk factor and the weight distribution result, a state transition probability matrix of the first cavity mode laser wavelength, the second cavity mode laser wavelength, the dual-cavity mode output interval duration and the dual-cavity mode laser output frequency and the welding fall-off probability is constructed, including: The welding log record data set is collected with the standard value of the welding fall-off risk factor as the limit, wherein the welding log record data set includes a first cavity mode laser wavelength record data set, a second cavity mode laser wavelength record data set, a dual-cavity mode output interval duration record data set, a dual-cavity mode laser output frequency record data set and a second welding fall-off identification data set, wherein the second welding fall-off identification data includes 0 or 1, 1 indicates that the welding has fallen off within the set service time, and 0 indicates that the welding has not fallen off within the set service time; Performing consistency clustering on the first cavity mode laser wavelength recording data set, the second cavity mode laser wavelength recording data set, the dual-cavity mode output interval duration recording data set, and the dual-cavity mode laser output frequency recording data set to obtain multiple groups of welding log recording data; Grouping the second welding peeling identification data according to the multiple groups of welding log record data to obtain multiple groups of second welding peeling identification data; Traversing the plurality of groups of second welding fall-off identification data, counting the proportion of the number of 0-value identifications, and obtaining a plurality of welding fall-off probabilities; Using the weight distribution result, weight distribution is performed on the multiple groups of welding log record data to construct multiple groups of welding control matrices; The state transition probability matrix is ​​constructed according to the multiple groups of welding control matrices and the multiple welding fall-off probabilities.

Images