CN113199184B - Weld joint shape prediction method based on improved self-adaptive fuzzy neural network - Google Patents
Weld joint shape prediction method based on improved self-adaptive fuzzy neural network Download PDFInfo
- Publication number
- CN113199184B CN113199184B CN202110754330.XA CN202110754330A CN113199184B CN 113199184 B CN113199184 B CN 113199184B CN 202110754330 A CN202110754330 A CN 202110754330A CN 113199184 B CN113199184 B CN 113199184B
- Authority
- CN
- China
- Prior art keywords
- welding
- fuzzy
- membership
- welding parameter
- neural network
- 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.)
- Active
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 71
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000003466 welding Methods 0.000 claims abstract description 309
- 230000003044 adaptive effect Effects 0.000 claims abstract description 16
- 239000002245 particle Substances 0.000 claims abstract description 16
- 238000004364 calculation method Methods 0.000 claims abstract description 15
- 238000005457 optimization Methods 0.000 claims abstract description 12
- 230000004927 fusion Effects 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 35
- 230000006870 function Effects 0.000 claims description 32
- 239000013598 vector Substances 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims 1
- 230000008569 process Effects 0.000 abstract description 22
- 238000012549 training Methods 0.000 abstract description 9
- 238000012360 testing method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 239000000463 material Substances 0.000 description 4
- 230000001681 protective effect Effects 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 229910001069 Ti alloy Inorganic materials 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K37/00—Auxiliary devices or processes, not specially adapted to a procedure covered by only one of the preceding main groups
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K37/00—Auxiliary devices or processes, not specially adapted to a procedure covered by only one of the preceding main groups
- B23K37/02—Carriages for supporting the welding or cutting element
- B23K37/0252—Steering means
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1602—Programme controls characterised by the control system, structure, architecture
- B25J9/161—Hardware, e.g. neural networks, fuzzy logic, interfaces, processor
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
- B25J9/1664—Programme controls characterised by programming, planning systems for manipulators characterised by motion, path, trajectory planning
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1679—Programme controls characterised by the tasks executed
-
- 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
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mechanical Engineering (AREA)
- General Physics & Mathematics (AREA)
- Robotics (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Optics & Photonics (AREA)
- Automation & Control Theory (AREA)
- Probability & Statistics with Applications (AREA)
- Fuzzy Systems (AREA)
- Feedback Control In General (AREA)
Abstract
The embodiment of the invention provides a weld joint shape prediction method based on an improved self-adaptive fuzzy neural network, which is used for initializing the weld joint shape prediction fuzzy neural network and determining a welding fuzzy rule; determining input welding parameter variables of the fuzzy neural network and output quantity of the fuzzy neural network; calculating the membership degree of the welding parameters of each welding parameter variable; optimizing and calculating unknown parameters in a welding parameter membership function by using an intuitive fuzzy C mean value clustering (IFCM) and adaptive inertial weight particle swarm optimization (APSO) fusion algorithm to obtain the welding parameter membership of each welding parameter variable; performing fuzzy calculation according to the membership degree of the welding parameters of each input variable; and predicting output values of the weld width and the weld height. The prediction method provided by the invention reduces the labor cost, effectively improves the accuracy of welding result prediction, and solves the problem that the traditional fuzzy neural network training process is easy to fall into local minimum points.
Description
Technical Field
The invention relates to the field of robot welding, in particular to a weld joint shape prediction method based on an improved self-adaptive fuzzy neural network, and relates to a robot welding weld joint shape prediction method.
Background
The quality of welding quality directly influences the processing precision and the service life of a welding part. In the welding process, the welding quality influence factors are numerous, and the welding quality detection of the titanium alloy mainly comprises the compactness, the physical property, the mechanical property, the welding defect, the appearance size and the like of a welding joint. The final welding quality is affected by abnormal fluctuation of welding parameters, vibration of the mechanical arm, selection of welding parameters by a welder and operation specifications of welding equipment. In the existing processing factory, welding workers generally set welding parameters and then sample test pieces are utilized to judge whether the parameter setting is reasonable or not, if the quality of the test pieces does not meet the requirements, the parameters are continuously adjusted and the sample test is carried out, a large amount of materials can be wasted in the process, and long time and labor are consumed in the process.
The existing neural network algorithm can map any complex nonlinear function due to the strong nonlinear capability, and is primarily applied to a robot welding system. However, in the actual welding process, the correlation between the welding parameters and the appearance characteristics of the welding seam is difficult to quantitatively describe, the factors influencing the welding quality of the robot are more, the fuzzy algorithm can express fuzzy knowledge, fuzzy reasoning is realized, and the defect of poor generalization capability of a neural network can be effectively overcome.
In 1984, a fuzzy C-means clustering algorithm (FCM) is proposed by J.C.Bezdek, membership degree of a sample to a clustering center is introduced, membership degree of each sample point to the clustering center is obtained by optimizing a target function, and the category of the sample points is determined so as to achieve the purpose of automatically classifying sample data. In 1995, Kennedy and Eberhart proposed a particle swarm optimization (POS) algorithm, which has the advantages of swarm intelligence, simple iteration format, rapid convergence to obtain the region where the optimal solution is located, and so on, and so far, both the FCM algorithm and the PSO algorithm have been applied well. However, in the process of determining the width of the membership function of the welding parameter by using the PSO algorithm, a smaller weight factor is beneficial to performing accurate local search on the current search area, a smaller weight factor is selected to solve the problem that the current search area is prone to fall into a local minimum point in the fuzzy neural network training process, a larger weight factor is selected to facilitate jumping out of the local minimum point, so that global search is facilitated, but the prediction accuracy is insufficient, the membership function is difficult to determine, and the prediction accuracy and the global optimality of the prediction result are difficult to take into account.
Disclosure of Invention
The invention provides a robot welding weld joint shape prediction method based on an improved self-adaptive fuzzy neural network, aiming at solving the fitting problem between welding parameters and weld joint outer dimension. According to the method, the welding seam appearance is predicted by adopting the self-adaptive fuzzy neural network, and the welding parameters are adjusted according to the predicted data.
A weld seam appearance prediction method based on an improved adaptive fuzzy neural network comprises the following steps:
step 1: initializing a weld contour prediction fuzzy neural network, and determining a welding fuzzy rule;
step 2: determining input welding parameter variables of the fuzzy neural network and output quantity of the fuzzy neural network;
and step 3: calculating the welding parameter membership of each welding parameter variable according to the welding fuzzy rule determined in the step 1;
and 4, step 4: optimizing and calculating unknown parameters in the welding parameter membership by using an intuitive fuzzy C mean value clustering (IFCM) and adaptive inertial weight particle swarm optimization (APSO) fusion algorithm to obtain the welding parameter membership of each welding parameter variable;
and 5: performing fuzzy calculation according to the welding parameter membership of each welding parameter variable;
step 6: obtaining the predicted output values of the width and height of the welding seam of the robot according to the fuzzy calculation result obtained in the step 5;
the step 3 comprises the following steps:
for input welding parameter variablesCalculating the welding parameter membership of each welding parameter input quantity according to the welding fuzzy rule determined in the step 1:
wherein,andrespectively the center and the width of the membership function of the welding parameters,is the degree of membership of the welding parameters,is based onInput quantity of welding parametersThe membership of the welding parameters calculated by the fuzzy subsets,is the fuzzy subset number of the welding parameter variables,is the number of welding parameter variables entered.
Preferably, the step 1 specifically includes:
describing the weld shape prediction fuzzy neural network by using an if-then definition rule, wherein the weld shape prediction fuzzy neural network can be defined by using the following if-then rule:
wherein,is the firstStrip welding dieThe rule is pasted,,is the number of weld fuzzy rules;is a variable of a welding parameterTo (1) aA fuzzy subset;is an input welding parameter variable, and;for the number of welding parameter variables that are input,is the firstOutput of the bar fuzzy rule, whereinAre the weight coefficients.
Preferably, the step 2 specifically includes:
the flow of shielding gas, the welding speed, the wire feeding speed and the laser power are taken as four input welding parameter variables of the fuzzy neural network, and the welding width and the welding height are respectively taken as network output.
Preferably, the step 4 specifically includes:
step 4.1, giving an initial value to the membership matrix of the welding parameter by using a random number generator;
step 4.2, determining a welding direct fuzzy set, calculating the uncertainty of a welding parameter sample based on the welding direct fuzzy set, and changing a membership matrix of welding parameters into a fuzzy membership matrix;
4.3, calculating the distance from the welding parameter to be classified to the clustering center based on the fuzzy membership matrix, and dividing the sample into various classes;
step 4.4, recalculating the distance from the clustering center of each class and the welding parameter sample to the clustering center, replacing the original membership matrix with an intuitive fuzzy membership matrix, and reclassifying the samples into each class;
step 4.5, iteratively searching the optimal solution of the membership function;
and 4.6, repeating the steps 4.2-4.5 until the fitness function reaches a specified threshold value.
Preferably, the step 4.2 specifically includes:
determining a welding intuition fuzzy set WIFS, and increasing the non-membership degree of welding parameter variables in the WIFSAnd uncertaintyAssuming intuitive fuzzy sets of weldingRepresenting welding parameter variablesAnd universe of discourseThe relationship of (1) is as follows:
wherein,for welding intuitive fuzzy setsThe degree of membership of (a) is,for welding intuitive fuzzy setsThe degree of non-membership of (a) is,for welding intuitive fuzzy setsWhen the condition is satisfiedAndand isThe uncertainty of the welding parameter variable may be expressed as:
preferably, the step 4.3 specifically includes:
defining the welding intuitive fuzzy membership as:
wherein,for welding intuitive fuzzy membership, expressA welding parameter point pairThe intuitive fuzzy membership of the clustering center of each welding parameter,is shown asA welding parameter variable is inThe uncertainty under each fuzzy subset,is shown asA welding parameter variable is inThe degree of membership under each fuzzy subset,,for the number of welding parameter variables that are input,for the number of fuzzy subsets of corresponding weld parameter variables,andcan be calculated from the following formula:
wherein,is the number of initialized membership matrix,for the number of welding parameter variables that are input,is a weight factor that is a function of,is a positive constant.
Preferably, the step 4.4 specifically includes:
by using an intuitive membership matrix formed by welding intuitive fuzzy membership of welding parameter variables, a new welding parameter clustering center formula can be obtained:
preferably, the step 5 specifically includes:
substituting the unknown number in the optimized welding parameter membership obtained in the step 4 into a calculation formula of the welding parameter membership, and calculating the membership of each input welding parameter in a fuzzy manner, wherein the fuzzy operator is a continuous multiplication operator:
preferably, the step 6 specifically includes:
and (5) substituting the fuzzy result obtained in the step (5) into an equation to calculate the predicted output values of the width and the height of the welding seam of the robot:
wherein,is a joint product of the degrees of membership of the input weld parameters, a weightThe following formula is adopted for updating:
in the formula,is a netThe welding seam welding width or the welding height output is expected;the actual welding seam width or welding height output of the network is realized;is the error between the desired output and the actual output.
The robot welding weld joint shape prediction method based on the improved adaptive fuzzy neural network realizes the prediction of the weld joint outer dimension based on the improved adaptive fuzzy neural network, takes the weld leg width and the weld height of a T-shaped weld joint as evaluation standards, selects four variables with the largest influence factors on welding quality as input parameters, optimizes the central value and the width of a membership function in the adaptive neural network fuzzy algorithm, and predicts the weld joint shape so as to adjust the welding parameters according to the prediction result. According to the robot welding seam appearance prediction method based on the improved self-adaptive fuzzy neural network, the fuzzy neural network is used for predicting the welding quality, so that the labor cost is reduced, and the material waste caused by manual trial welding is avoided; the input parameters and the output parameters of welding are subjected to nonlinear fitting, the accuracy of welding result prediction is effectively improved, two unknown parameters in a membership function are subjected to optimization calculation by using an IFCM-APSO algorithm, and the problem that a local minimum point is easy to fall into in the traditional fuzzy neural network training process is solved.
Drawings
FIG. 1 is a schematic view of a robotic welding platform;
FIG. 2 is a flowchart of a robot welding seam profile prediction method based on a fuzzy neural network in an embodiment;
FIG. 3 is a graph of the optimized membership function;
FIG. 4 is a shape parameter of a typical T-weld;
FIG. 5 is a comparison graph of weld profile prediction results before optimization in the examples;
FIG. 6 is a schematic diagram of a weld joint shape prediction result after optimization in the embodiment;
FIG. 7 is a welding result of a prior art trial welding method based on empirical search for welding parameters;
FIG. 8 is a diagram illustrating a welding result predicted by a robot welding seam profile prediction method based on a fuzzy neural network in an embodiment;
FIG. 9 is a flowchart of a training process of a neural network in an embodiment;
fig. 10 is a schematic diagram of a robotic welding station assembly.
Detailed Description
The technical solutions in the embodiments of the present application are described below clearly and completely with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. As can be known to those skilled in the art, with the development of technology and the emergence of new scenarios, the technical solution provided in the embodiments of the present application is also applicable to similar technical problems.
The terms "comprises" and "comprising," and any variations thereof, in the description and claims of this application and the above-described drawings are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of elements is not necessarily limited to those elements, but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
This example was carried out in a welding center using a robotic welding platform (platform assembly shown in fig. 1). The platform includes: welding protective gas, silk laser sensor, robot control cabinet, send a machine and laser instrument, KUKA arm and robot control cabinet signal connection, PLC switch board connects above-mentioned subassembly, realizes predictive computation, parameter setting and mechanical arm control.
Example one
The first embodiment of the invention relates to a shape prediction method based on an improved adaptive fuzzy neural network, the flow of which is shown in fig. 2, and the specific steps are as follows:
step 1: initializing a weld contour prediction fuzzy neural network, and determining a welding fuzzy rule;
step 2: determining input welding parameter variables of the fuzzy neural network and output quantity of the fuzzy neural network;
and step 3: calculating the welding parameter membership of each welding parameter variable according to the welding fuzzy rule determined in the step 1;
and 4, step 4: optimizing and calculating unknown parameters in the welding parameter membership by using an intuitive fuzzy C mean value clustering (IFCM) and adaptive inertial weight particle swarm optimization (APSO) fusion algorithm to obtain the welding parameter membership of each welding parameter variable;
and 5: performing fuzzy calculation according to the welding parameter membership of each welding parameter variable;
step 6: and 5, obtaining the predicted output values of the width and height of the welding seam of the robot according to the fuzzy calculation result obtained in the step 5.
According to the robot welding seam shape prediction method based on the improved self-adaptive fuzzy neural network, the nonlinear relation between the welding parameter variable and the neural network output quantity is obtained based on fuzzy neural network fitting, the width and the height of the welding seam are predicted according to the fuzzy neural network, compared with a traditional trial welding method, the shape after welding can be predicted according to the input welding parameters, the labor cost is reduced, and material waste caused by manual trial welding is avoided. The IFCM-APSO algorithm is used for carrying out optimization calculation on two unknown parameters in the membership function, and the problem that a local minimum point is easy to fall into in the traditional fuzzy neural network training process is solved.
Example two
Further, a second embodiment of the present invention relates to a weld seam shape prediction method based on an improved adaptive fuzzy neural network, which specifically includes the following steps:
step 1: initializing a weld contour prediction fuzzy neural network, and determining a welding fuzzy rule;
describing the weld shape prediction fuzzy neural network by using an if-then definition rule, wherein the weld shape prediction fuzzy neural network in the step 1 can be defined by using the following if-then rule:
wherein,is the firstThe rule of the welding of the bars is fuzzy,,is the number of weld fuzzy rules;is a variable of a welding parameterTo (1) aA fuzzy subset;is an input welding parameter variable, and;for the number of welding parameter variables that are input,is the firstOutput of the bar fuzzy rule, whereinAre the weight coefficients.
Step 2: determining input welding parameter variables of the fuzzy neural network and output quantity of the fuzzy neural network;
in the welding process, factors affecting the quality of a welded product are numerous. When the flow of the protective gas is too large, the gas impacts a molten pool, so that the splashing of the molten pool is increased, and the surface of a welding seam is not smooth; when the flow of the protective gas is too small, the protective effect on a molten pool is reduced, and defects such as air holes and the like are easily generated. When the welding speed is too high, the gas protection effect is damaged, so that the welding seam is not formed well; when the welding speed is too slow, the fusion width is too large, the molten pool becomes large, and the material is easy to weld through. If the wire feeding speed is too high, the welding width and the welding height of a welding seam can be increased; if the speed is too low, cold joint is easily caused, and the mechanical property of a welding workpiece is influenced. The selection of laser power also has great influence on the formation of a welding seam, the melting depth is increased quickly when the power is too large, and the welding width and the welding height are correspondingly increased. Therefore, the flow of the shielding gas, the welding speed, the wire feeding speed and the laser power are taken as welding parameter variables of four inputs of the fuzzy neural network, and the welding width and the welding height are taken as network outputs respectively.
And step 3: calculating the welding parameter membership of each welding parameter variable according to the welding fuzzy rule determined in the step 1;
in step 3, for the input welding parameter variableCalculating the welding parameter membership of each welding parameter input quantity according to the welding fuzzy rule determined in the step 1:
wherein,andrespectively the center and the width of the membership function of the welding parameters,is the degree of membership of the welding parameters,is based onInput quantity of welding parametersThe membership of the welding parameters calculated by the fuzzy subsets,is the fuzzy subset number of the welding parameter variables,is the number of welding parameter variables entered.
And 4, step 4: optimizing and calculating unknown parameters in the welding parameter membership by using an intuitive fuzzy C mean value clustering (IFCM) and adaptive inertial weight particle swarm optimization (APSO) fusion algorithm to obtain the welding parameter membership of each welding parameter variable;
step 4.1, giving an initial value to the membership matrix of the welding parameter by using a random number generator;
and the membership matrix of the welding parameters is formed by the membership of the welding parameters of all welding parameter variables.
Step 4.2, determining a welding direct fuzzy set, calculating the uncertainty of a welding parameter sample based on the welding direct fuzzy set, and changing a membership matrix of welding parameters into a fuzzy membership matrix;
firstly, determining a welding intuitive modelThe method is characterized in that a fuzzy set WIFS is an important expansion of a traditional fuzzy set, and the non-membership degree of welding parameter variables is increased in the WIFSAnd uncertaintyAssuming intuitive fuzzy sets of weldingRepresenting welding parameter variablesAnd universe of discourseThe relationship of (1) is as follows:
wherein,for welding intuitive fuzzy setsThe degree of membership of (a) is,for welding intuitive fuzzy setsThe degree of non-membership of (a) is,for welding intuitive fuzzy setsUncertainty of (d). When satisfying the stripPieceAndand isThe uncertainty of the welding parameter variable may be expressed as:
4.3, calculating the distance from the welding parameter to be classified to the clustering center based on the fuzzy membership matrix, and dividing the sample into various classes;
in order to combine the welding intuitive fuzzy characteristic with the traditional fuzzy clustering method, the welding intuitive fuzzy membership is defined as:
wherein,for welding intuitive fuzzy membership, expressA welding parameter point pairThe intuitive fuzzy membership of the clustering center of each welding parameter,is shown asA welding parameter variable is inThe uncertainty under each fuzzy subset,is shown asA welding parameter variable is inThe degree of membership under each fuzzy subset,,for the number of welding parameter variables that are input,the number of fuzzy subsets of the corresponding welding parameter variables.Andcan be calculated from the following formula:
wherein,is the number of initialized membership matrix,for the number of welding parameter variables that are input,is a weight factor that is a function of,a positive constant to ensure that the sum of the degree of membership and the degree of non-membership is no greater than 1.
And 4.4, recalculating the distance from the cluster center of each class and the welding parameter sample to the cluster center. Each calculation uses an intuitive fuzzy membership matrix to replace the original membership matrix, and samples are divided into various classes again;
by using an intuitive membership matrix formed by welding intuitive fuzzy membership of welding parameter variables, a new welding parameter clustering center formula can be obtained:
4.5, iterating by using the objective function, and searching an optimal solution of the membership function;
the IFCM-APSO objective function can be expressed as:
wherein,,is shown asThe data of the 4-dimensional vectors,is shown asCommon in each clusterThe number of the elements is one,is the number of the centers of the clusters,is as followsThe number of data points is, for example,is as followsThe center of each cluster is determined by the center of each cluster,is shown asIndividual clustering point pairIntuitive fuzzy membership of individual cluster centers;as a weighting factor, in general;For distortion, it is generally expressed as Euclidean distance;In order to be the true value of the value,predicting an output value for the neural network; FIG. 3 is the optimized membership function parameters.
And 4.6, repeating the steps 4.2-4.5 until the fitness function reaches a specified threshold value.
And updating the intuitive membership matrix of the welding parameters by using the updated value of the welding parameter clustering center. In the process of each iteration, the numerical value of the welding parameter intuitive membership matrix of the welding parameter clustering center is updated once until the difference value between the former welding parameter intuitive membership matrix and the updated welding parameter intuitive membership matrix is smaller than a set threshold value, at the moment, the iteration process is finished, and the clustering center reaches the optimum value.
The fitness function is as follows:
the above formula is an iterative objective function formula of IFCM, and the membership matrix after parameter updatingAnd membership matrix before updateIs less thanIf so, ending the iteration process; finding the optimum width of the membership function using modified particle swarm optimization, i.e.A value of (d); particlesAre grouped intoPopulation of individual particles
The speed is updated to,Is a local learning factor and is used as a local learning factor,in order to be a global learning factor,to representAt the first momentThe velocity of the individual particles is determined,represents the historical best record of the individual particle,indicating the optimal position for the global duration,to representAt the first momentThe position of the particles is determined by the position of the particles,andrandom number of 0 to 1, position update to
wherein,,respectively representThe maximum value and the minimum value of (c),the current value of the objective function for the particle is indicated,andrespectively representing the average target value and the minimum target value of all the particles at present.
And 5: performing fuzzy calculation according to the welding parameter membership of each welding parameter variable;
substituting the unknown number in the optimized welding parameter membership obtained in the step 4 into a calculation formula of the welding parameter membership, and calculating the membership of each input welding parameter in a fuzzy manner, wherein the fuzzy operator is a continuous multiplication operator:
step 6: and 5, obtaining the predicted output values of the width and height of the welding seam of the robot according to the fuzzy calculation result obtained in the step 5.
And (5) substituting the fuzzy result obtained in the step (5) into an equation to calculate the predicted output values of the width and the height of the welding seam of the robot:
wherein,is a joint product of the degrees of membership of the input weld parameters, a weightThe following formula is adopted for updating:
in the formula,is the expected weld width or weld height output of the network;the actual welding seam width or welding height output of the network is realized;is the error between the desired output and the actual output.
According to the robot welding seam appearance prediction method based on the improved adaptive fuzzy neural network, four input welding parameters with large influence factors are selected to respectively predict the welding width and the welding height, in the prediction process, the prediction is carried out based on the improved adaptive fuzzy neural network, the input and output parameters of the welding are subjected to nonlinear fitting, and the accuracy of welding result prediction is effectively improved; the local learning factor and the global learning factor are simultaneously considered in the speed updating, and the problem that the traditional fuzzy neural network training process is easy to fall into a local minimum point is solved.
EXAMPLE III
The accuracy and applicability of the fuzzy neural network depend to a large extent on the available training data set, the larger the input range it covers, and the better the network generated. In order to compare the predicted performance of the algorithm, the invention carries out welding test on the titanium alloy steel plate with the thickness of 3mm, and generates 250 data sets by changing the process parameters. Randomly selecting 200 of the fuzzy neural networks for training, and testing the fitting degree of the networks for the rest 50; two independent fuzzy neural network models are established and are respectively used for predicting welding width and welding height data. The shape parameters of a typical T-weld are shown in fig. 4.
The fuzzy neural network determines that the number of input nodes is 4 and the number of output nodes is 1 according to the input and output dimensions of the training sample, and artificially determines that the number of membership function is 8 according to the number of input and output nodes of the network. In order to calculate the central value and the width of the membership function, an IFCM-APSO algorithm programming program is adopted in Matlab2019b to optimize the parameters of the predecessor, and data testing is performed on the optimized fuzzy neural network, wherein the error change before and after optimization is shown in fig. 5 and 6.
Compared with actual output, the predicted output data has larger fluctuation, after the front-part parameters are optimized by the IFCM-APSO algorithm, the fitting degree of the fuzzy neural network is greatly improved, the test data set is substituted into the optimized fuzzy neural network, and the output predicted data can reduce the error between the predicted output value and the actual output value.
FIG. 7 shows the welding results obtained after five parameter adjustments by using the manual trial welding method, and it can be seen that the weld surface is rough and uneven, and the expected welding effect is difficult to achieve within the range of the limited number of adjustments; and FIG. 8 is a weld forming diagram obtained by using the improved fuzzy neural network to adjust the welding parameters three times before welding according to the prediction result, and the welding effect diagram shows that the improved fuzzy neural network can predict the overall dimension of the weld within a certain error range, thereby effectively improving the adjustment efficiency of the welding parameters.
The steps of the above methods are divided for clarity, and the implementation may be combined into one step or split some steps, and the steps are divided into multiple steps, so long as the same logical relationship is included, which are all within the protection scope of the present patent; it is within the scope of the patent to add insignificant modifications to the algorithms or processes or to introduce insignificant design changes to the core design without changing the algorithms or processes.
The above examples are only specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can modify or change the above embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (4)
1. A weld seam appearance prediction method based on an improved adaptive fuzzy neural network is characterized by comprising the following steps:
step 1: initializing a weld contour prediction fuzzy neural network, and determining a welding fuzzy rule;
step 2: determining input welding parameter variables of the fuzzy neural network and output quantity of the fuzzy neural network;
and step 3: calculating the welding parameter membership of each welding parameter variable according to the welding fuzzy rule determined in the step 1;
and 4, step 4: optimizing and calculating unknown parameters in the welding parameter membership by using a fusion algorithm based on intuitive fuzzy C-means clustering and self-adaptive inertial weight particle swarm optimization to obtain the welding parameter membership of each welding parameter variable;
and 5: performing fuzzy calculation according to the welding parameter membership of each welding parameter variable;
step 6: obtaining the predicted output values of the width and height of the welding seam of the robot according to the fuzzy calculation result obtained in the step 5;
the step 3 comprises the following steps:
for input welding parameter variablesCalculating the welding parameter membership of each welding parameter input quantity according to the welding fuzzy rule determined in the step 1:
wherein,andrespectively the center and the width of the membership function of the welding parameters,is the degree of membership of the welding parameters,is based onInput quantity of welding parametersThe membership of the welding parameters calculated by the fuzzy subsets,is the fuzzy subset number of the welding parameter variables,is the number of input welding parameter variables;
the step 4 specifically includes:
step 4.1, assigning an initial value to a membership matrix of the welding parameters by using a random number generator, wherein the membership matrix of the welding parameters is formed by the membership of the welding parameters of all welding parameter variables;
step 4.2, determining a welding intuitive fuzzy set, calculating the uncertainty of a welding parameter sample based on the welding intuitive fuzzy set, and changing a membership matrix of welding parameters into a fuzzy membership matrix;
determining a welding intuition fuzzy set WIFS, and increasing the non-membership degree of welding parameter variables in the WIFSAnd uncertaintyAssuming intuitive fuzzy sets of weldingRepresenting welding parameter variablesAnd universe of discourseThe relationship of (1) is as follows:
wherein,intuitive fuzzy set for weldingThe degree of membership of (a) is,intuitive fuzzy set for weldingThe degree of non-membership of (a) is,intuitive fuzzy set for weldingWhen the condition is satisfiedAndand isThe uncertainty of the welding parameter variable may be expressed as:
4.3, calculating the distance from the welding parameter to be classified to the clustering center based on the fuzzy membership matrix, and dividing the sample into various classes;
wherein, the welding intuitive fuzzy membership is defined as:
wherein,is shown asA welding parameter point pairThe intuitive fuzzy membership of the clustering center of each welding parameter,is shown asA welding parameter variable is inThe uncertainty under each fuzzy subset,is shown asA welding parameter variable is inThe degree of membership under each fuzzy subset,,for the number of welding parameter variables that are input,for corresponding welding parameter variablesThe number of the fuzzy subsets is determined,
wherein,
is the number of initialized membership matrix,for the number of welding parameter variables that are input,is a weight factor that is a function of,is a positive constant;
step 4.4, recalculating the distance from the clustering center of each class and the welding parameter sample to the clustering center, replacing the original membership matrix with an intuitive fuzzy membership matrix, and reclassifying the samples into each class;
obtaining a new welding parameter clustering center formula by using an intuitive membership matrix formed by welding intuitive fuzzy membership of welding parameter variables:
step 4.5, iteratively searching the optimal solution of the welding parameter membership;
performing iteration by using an objective function, wherein the objective function is as follows:
wherein,,is shown asThe data of the 4-dimensional vectors,is shown asCommon in each clusterThe number of the elements is one,is the number of the centers of the clusters,is as followsThe number of data points is, for example,is as followsThe center of each cluster is determined by the center of each cluster,is shown asIndividual clustering point pairIntuitive fuzzy membership of individual cluster centers;is a weight factor;expressed by Euclidean distance for distortion degree;In order to be the true value of the value,predicting an output value for the neural network;
step 4.6, repeating the steps 4.2-4.5 until the fitness function reaches a specified threshold value;
updating the welding parameter intuitive membership matrix by using the updated value of the welding parameter clustering center, and if the difference value of the former welding parameter intuitive membership matrix and the updated welding parameter intuitive membership matrix is less than a set threshold valueThe iterative process is ended;
wherein the fitness function is as follows:
the step 5 specifically includes:
substituting the unknown number in the optimized welding parameter membership obtained in the step 4 into a calculation formula of the welding parameter membership, and calculating the membership of each input welding parameter in a fuzzy manner, wherein the fuzzy operator is a continuous multiplication operator:
2. the weld joint shape prediction method based on the improved adaptive fuzzy neural network according to claim 1, wherein the step 1 specifically comprises:
describing the weld shape prediction fuzzy neural network by using an if-then definition rule, wherein the weld shape prediction fuzzy neural network can be defined by using the following if-then rule:
wherein,is the firstThe rule of the welding of the bars is fuzzy,,is the number of weld fuzzy rules;is a variable of a welding parameterTo (1) aA fuzzy subset;is an input welding parameter variable, and;for the number of welding parameter variables that are input,is the firstOutput of the bar fuzzy rule, whereinAre the weight coefficients.
3. The weld seam shape prediction method based on the improved adaptive fuzzy neural network according to claim 1, wherein the step 2 specifically comprises:
the flow of shielding gas, the welding speed, the wire feeding speed and the laser power are taken as four input welding parameter variables of the fuzzy neural network, and the welding width and the welding height are respectively taken as network output.
4. The weld seam shape prediction method based on the improved adaptive fuzzy neural network according to claim 1, wherein the step 6 specifically comprises:
and (5) substituting the fuzzy result obtained in the step (5) into an equation to calculate the predicted output values of the width and the height of the welding seam of the robot:
wherein,is a joint product of the degrees of membership of the input weld parameters, a weightThe following formula is adopted for updating:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110754330.XA CN113199184B (en) | 2021-07-05 | 2021-07-05 | Weld joint shape prediction method based on improved self-adaptive fuzzy neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110754330.XA CN113199184B (en) | 2021-07-05 | 2021-07-05 | Weld joint shape prediction method based on improved self-adaptive fuzzy neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113199184A CN113199184A (en) | 2021-08-03 |
CN113199184B true CN113199184B (en) | 2021-10-08 |
Family
ID=77022649
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110754330.XA Active CN113199184B (en) | 2021-07-05 | 2021-07-05 | Weld joint shape prediction method based on improved self-adaptive fuzzy neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113199184B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113536489B (en) * | 2021-08-25 | 2023-06-20 | 中国电子科技集团公司第三十八研究所 | Method for determining connection configuration and process parameters of component package |
CN115609180B (en) * | 2022-12-01 | 2023-03-10 | 江苏新恒基特种装备股份有限公司 | Strengthening parameter generation method and system based on welding process prediction and storage medium |
CN117548824B (en) * | 2024-01-11 | 2024-04-02 | 武汉新耐视智能科技有限责任公司 | Laser remote welding method for optical precision measurement robot |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR100373140B1 (en) * | 2001-02-09 | 2003-02-25 | 현대중공업 주식회사 | Automatic seam tracking system in conjunction with quality control by using artificial intelligence |
CN1546268A (en) * | 2003-12-17 | 2004-11-17 | 南昌大学 | Intelligent control system for weld seam tracking and fusion penetration in spiral pipes |
CN203156243U (en) * | 2012-12-17 | 2013-08-28 | 海目(北京)激光与数控发展有限公司 | Automatic protection equipment for laser conduction device and laser |
CN103264216B (en) * | 2013-05-15 | 2016-05-04 | 山东大学 | Based on controlled plasma arc welding system and the technique of back side aperture vision-based detection |
CN103521890B (en) * | 2013-10-12 | 2015-09-30 | 王晓宇 | Two-sided pair of arc vertical position welding near infrared vision sensing and penetration control device and method |
CN107598401A (en) * | 2016-07-12 | 2018-01-19 | 沈阳富创精密设备有限公司 | Plasma arc welding (PAW) electric current molten wide control system based on Fuzzy Reasoning Neural Network |
-
2021
- 2021-07-05 CN CN202110754330.XA patent/CN113199184B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN113199184A (en) | 2021-08-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113199184B (en) | Weld joint shape prediction method based on improved self-adaptive fuzzy neural network | |
Zuperl et al. | Optimization of cutting conditions during cutting by using neural networks | |
JP4681426B2 (en) | Apparatus and method for analyzing relation between operation and quality in manufacturing process, computer program, and computer-readable recording medium | |
US10442023B2 (en) | Simulation apparatus of wire electric discharge machine having function of determining welding positions of core using machine learning | |
US5485390A (en) | Inductive-deductive process design for machined parts | |
Savkovic et al. | Comparison of artificial neural network, fuzzy logic and genetic algorithm for cutting temperature and surface roughness prediction during the face milling process | |
Wang | Multiple-objective optimisation of machining operations based on neural networks | |
JP5068637B2 (en) | Operation and quality related analysis apparatus, analysis method, program, and computer-readable recording medium in manufacturing process | |
Bloch et al. | Neural intelligent control for a steel plant | |
Tseng | Welding parameters optimization for economic design using neural approximation and genetic algorithm | |
CN113128124B (en) | Multi-grade C-Mn steel mechanical property prediction method based on improved neural network | |
CN114492198A (en) | Cutting force prediction method based on improved PSO algorithm assisted SVM algorithm | |
CN111651929A (en) | Multi-objective optimization method based on fusion of Dynaform and intelligent algorithm | |
CN113642220A (en) | Ship welding process optimization method based on RBF and MOPSO | |
Cica et al. | Intelligent process modeling and optimization of porosity formation in high-pressure die casting | |
JP4653547B2 (en) | Apparatus and method for analyzing relation between operation and quality in manufacturing process, computer program, and computer-readable recording medium | |
CN110837857A (en) | Industrial electricity load prediction method, system and storage medium thereof | |
CN116842814A (en) | Ship GMAW weld joint forming prediction method based on particle filtering | |
Selvakumar et al. | A novel approach for optimization to verify RSM model by using multi-objective genetic algorithm (MOGA) | |
JP3056324B2 (en) | Learning / recalling method and device | |
Strasser et al. | An Approach for Adaptive Parameter Setting in Manufacturing Processes. | |
Hallmann et al. | A method for analyzing the influence of process and design parameters on the build time of additively manufactured components | |
JP2021186883A (en) | Processing condition recommendation device, processing condition recommendation method, program, metal structure manufacturing system and metal structure manufacturing method | |
CN110814342B (en) | Preparation method of computer-formalized 3D printing metal material | |
CN114092472A (en) | Method, device and medium for detecting uncertain samples in defect detection |
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 |