Welding problem tracing method and device, computer equipment and storage medium

Abstract

The invention relates to a welding problem tracing method and device, computer equipment and a storage medium. The method comprises the following steps: performing data acquisition on process data, service information system data and quality manufacturing data of a terminal industrial personal computer, and performing data rough classification according to relevancy of the data and process quality to form rough classification data; performing data cleaning and data standardization on the rough classification data to obtain various welding process data sets and storing the welding process data sets in a database; selecting process parameters according to the various welding process data sets to construct a welding problem traceability model, and training the welding problem traceability model based on a welding process quality traceability process; and the trained welding problem traceability model is utilized to carry out influence factor positioning and sorting, traceability is carried out to obtain process parameters generated by unqualified welding products and the proportion of the influence factors, and the process parameters and the proportion are output. According to the method, efficient and accurate prediction can be provided for the welding process quality problem, and the traceability precision of the welding process quality problem can be improved.

Description

Technical Field

[0001] The present application relates to the field of welding process technology, and in particular to a welding problem tracing method, device, computer equipment and storage medium. Background Art

[0002] Welding technology is the mainstream processing technology in the production process of ship hulls, and the welding quality will have a great impact on the quality of ships. In the actual hull manufacturing process, the welding process parameters will be adjusted according to multiple process tests, and finally a set of welding process parameters that are most suitable for production will be selected. Most welding process tests rely on the experience of welders to adjust parameters, which is time-consuming, inefficient, and has a high repetition rate. Most current technologies use radial basis function neural networks (RBF neural networks for short) to establish an inference model for welding process parameters, but this technology has low traceability accuracy for welding process quality problems under large amounts of data and large-scale networks, and cannot provide efficient and accurate predictions. Therefore, a method based on a back-propagation neural network model is established to trace the welding process parameters, and the problem process parameters can be located with high precision in a large number of data sets, which can effectively reduce the number of welding process tests and reduce production costs. Summary of the invention

[0003] Based on this, a welding problem tracing method, device, computer equipment and storage medium are provided to solve the technical problem that the tracing accuracy of welding process quality problems is not high and efficient and accurate prediction cannot be provided.

[0004] On the one hand, a welding problem tracing method is provided, the method comprising: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

[0005] In one embodiment, the terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; and the quality manufacturing data includes quality manufacturing data extracted from a hull section manufacturing quality database and a defective product quality database.

[0006] In one embodiment, the data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

[0007] In one embodiment, selecting process parameters according to the various welding process data sets to construct a welding problem tracing model includes: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

[0008] In one embodiment, the welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

[0009] In one embodiment, pre-training and fine-tuning the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

[0010] In one embodiment, the trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

[0011] On the other hand, a welding problem tracing device is provided, the device comprising: The data rough classification module is used to collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and to roughly classify the data according to the correlation between the data and the process quality to form rough classification data; A welding process data set processing module is used to perform data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and store them in a database; A model building and training module, used to select process parameters according to the various welding process data sets to build a welding problem tracing model, and to train the welding problem tracing model based on the welding process quality tracing process; The model application module is used to use the trained welding problem tracing model to locate and sort the influencing factors, trace the process parameters of welding defective products and the proportion of their influencing factors, and output them.

[0012] In another aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are implemented: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

[0013] In another aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

[0014] The above-mentioned welding problem tracing method, device, computer equipment and storage medium obtain various types of welding process data sets by performing data cleaning and data standardization on coarse classification data, and select process parameters according to the various types of welding process data sets to construct a welding problem tracing model, wherein the welding problem tracing model is based on welding process quality tracing process training, can effectively locate and sort influencing factors, trace back to obtain process parameters and the proportion of influencing factors that produce defective welding products, can provide efficient and accurate prediction of welding process quality problems, and can also improve the tracing accuracy of welding process quality problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 This is a flow chart of a welding problem tracing method in one embodiment of the present application; Figure 2 This is a flowchart of data set processing in one embodiment of the present application; Figure 3 A schematic diagram of a process for constructing a welding problem tracing model in one embodiment of the present application; Figure 4 This is a schematic diagram of the structure of a deep belief network in one embodiment of the present application; Figure 5 This is a flow chart of training the welding problem tracing model based on the welding process quality tracing process in one embodiment of the present application; Figure 6 A logic diagram for training the welding problem tracing model in one embodiment of the present application; Figure 7 This is a structural block diagram of a welding problem tracing device in one embodiment of the present application; Figure 8 This is a diagram of the internal structure of a computer device in one embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0018] In one embodiment, Figure 1 As shown, a welding problem tracing method is provided, comprising the following steps: Step S1, collecting the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classifying the data according to the relevance between the data and the process quality to form roughly classified data; Step S2, performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and store them in a database; Step S3, selecting process parameters according to the various welding process data sets to build a welding problem tracing model, and training the welding problem tracing model based on the welding process quality tracing process; Step S4, using the trained welding problem tracing model to locate and sort the influencing factors, tracing back to obtain the process parameters and the proportion of influencing factors that cause welding defective products and output them.

[0019] Among them, various types of welding process data sets are obtained by cleaning and standardizing the coarse classification data, and process parameters are selected according to the various types of welding process data sets to build a welding problem tracing model. The welding problem tracing model is based on the welding process quality traceability process training, which can effectively locate and sort the influencing factors, and trace the process parameters and the proportion of influencing factors that cause welding defectives. It can provide efficient and accurate prediction of welding process quality problems, and can also improve the traceability accuracy of welding process quality problems.

[0020] In this embodiment, the terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; the quality manufacturing data includes quality manufacturing data extracted from the hull section manufacturing quality database and the defective product quality database.

[0021] In this embodiment, the data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

[0022] See also Figure 2 After the coarse classification data is cleaned and standardized, it also includes Trace the process parameters, sort out the influencing factors, and adjust the process parameters.

[0023] See also Figure 2 , Figure 3 In this embodiment, selecting process parameters according to the various welding process data sets to construct a welding problem tracing model includes: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

[0024] Among them, the structural construction of the welding problem tracing model mainly refers to the use of Deep Belief Networks (DBNs) for structural construction. DBNs is a generative model composed of multiple layers of Restricted Boltzmann Machines (RBMs). Restricted Boltzmann Machine (RBM) is the basic building block of deep belief networks. RBM is a generative random neural network consisting of two layers of fully connected neurons: a visible layer and a hidden layer. The visible layer contains neurons that directly encode the data, and the hidden layer contains neurons that learn features from the visible layer.

[0025] The working principle of RBM is based on an energy function, which defines the energy of the state of the network. The energy of an RBM is related to the joint probability distribution of its states, where lower energy corresponds to higher probability.

[0026] The RBM algorithm consists of the following main steps: forward propagation, activation from visible layers to hidden layers. Backward propagation, reconstruction from hidden layers to visible layers. Gradient calculation, gradient of weight update is calculated by contrastive divergence. Weight update, weight is updated by learning rate.

[0027] like Figure 4 As shown in Figure 1, the structure of DBN consists of multiple layers, usually including multiple restricted Boltzmann machine (RBM) layers and a top layer. Each layer consists of a group of neurons connected to the neurons of the adjacent layer through bidirectional connections.

[0028] The input layer corresponds to the visible representation of the data. The hidden layer includes multiple RBM layers, each of which corresponds to a higher level of abstraction of the data. The top layer, usually composed of an RBM or other model, is responsible for the extraction and representation of the final features, serving as the output layer.

[0029] The connection structure of DBNs follows the following rules: there are no connections between neurons in the same layer, neurons in each layer are connected to all neurons in the upper and lower layers, and the connections are undirected (for the first few layers of RBM) or directed (for the top layer).

[0030] Among them, the combination of welding process parameters that is most likely to affect the actual process is: welding current, welding voltage, laser power, welding speed and wire feeding speed. The process parameter adjustment refers to the calculation of the corresponding process parameters according to the algorithm. For example, there is a 70% probability that the laser power is too large. After sorting the process parameters, the one with the largest output ratio is selected, and the corresponding adjustment suggestion is given to reduce it.

[0031] like Figure 5 As shown, in this embodiment, the welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

[0032] Among them, DBNs are used to train the welding problem tracing model. First, in the pre-training stage, each RBM layer is greedily trained layer by layer in order from bottom to top. Then, in the fine-tuning stage, the entire network is fine-tuned using the supervised learning method of back propagation; pre-training is the first stage of DBN training, which initializes the network weights. First, each RBM layer of DBN is trained separately, layer by layer from bottom to top. Then, the RBM is trained using the contrastive divergence algorithm. After each layer is trained, its weight is used as the input of the next layer. Fine-tuning is the second stage of DBN training, which adjusts the pre-trained weights to improve performance. First, supervised learning is performed using the back propagation algorithm. Then, the prediction error of the training data is minimized by adjusting the weights. During the process, it is important to prevent overfitting by monitoring the performance on the validation set.

[0033] like Figure 6 As shown, in this embodiment, the pre-training and fine-tuning of the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

[0034] In this embodiment, the trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

[0035] In the above-mentioned welding problem tracing method, various welding process data sets are obtained by performing data cleaning and data standardization on the coarse classification data, and process parameters are selected according to the various welding process data sets to construct a welding problem tracing model, wherein the welding problem tracing model is based on the welding process quality tracing process training, can effectively locate and sort the influencing factors, and trace back to obtain the process parameters and the proportion of influencing factors that produce defective welding products, which can provide efficient and accurate prediction of welding process quality problems, and can also improve the tracing accuracy of welding process quality problems.

[0036] In one embodiment, Figure 7 As shown, a welding problem tracing device 10 is provided, comprising: a data rough classification module 1, a welding process data set processing module 2, a model building and training module 3, and a model application module 4.

[0037] The data rough classification module 1 is used to collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and to roughly classify the data according to the relevance between the data and the process quality to form rough classification data.

[0038] The welding process data set processing module 2 is used to perform data cleaning and data standardization on the coarse classification data, obtain various types of welding process data sets and store them in a database.

[0039] The model building and training module 3 is used to select process parameters according to the various welding process data sets to build a welding problem tracing model, and to train the welding problem tracing model based on the welding process quality tracing process.

[0040] The model application module 4 is used to locate and sort the influencing factors using the trained welding problem tracing model, trace the process parameters of welding defective products and the proportion of their influencing factors, and output them.

[0041] In this embodiment, the terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; the quality manufacturing data includes quality manufacturing data extracted from the hull section manufacturing quality database and the defective product quality database.

[0042] In this embodiment, the data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

[0043] In this embodiment, selecting process parameters according to the various welding process data sets to construct a welding problem tracing model includes: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

[0044] In this embodiment, the welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

[0045] In this embodiment, pre-training and fine-tuning the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

[0046] In this embodiment, the trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

[0047] In the above-mentioned welding problem tracing device, various welding process data sets are obtained by cleaning and standardizing the coarse classification data, and process parameters are selected according to the various welding process data sets to construct a welding problem tracing model, wherein the welding problem tracing model is based on the welding process quality tracing process training, can effectively locate and sort the influencing factors, and trace the process parameters and the proportion of influencing factors that produce defective welding products, which can provide efficient and accurate prediction of welding process quality problems and improve the tracing accuracy of welding process quality problems.

[0048] The specific definition of the welding problem tracing device can be found in the definition of the welding problem tracing method mentioned above, which will not be repeated here. Each module in the above welding problem tracing device can be implemented in whole or in part by software, hardware and their combination. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0049] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

[0050] In one embodiment, the computer program further implements the following steps when executed by a processor: The terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; the quality manufacturing data includes the quality manufacturing data extracted from the hull section manufacturing quality database and the defective product quality database.

[0051] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

[0052] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The selecting process parameters according to the various welding process data sets to construct a welding problem tracing model includes: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

[0053] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

[0054] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The pre-training and fine-tuning of the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

[0055] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

[0056] For the specific limitations on the steps implemented when the computer program is executed by the processor, please refer to the limitations on the method for tracing the source of welding problems mentioned above, which will not be repeated here.

[0057] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store welding problem tracing data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a welding problem tracing method is implemented.

[0058] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0059] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

[0060] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: The terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; the quality manufacturing data includes the quality manufacturing data extracted from the hull section manufacturing quality database and the defective product quality database.

[0061] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: The data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

[0062] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: The selecting process parameters according to the various welding process data sets to construct a welding problem tracing model includes: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

[0063] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: The welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

[0064] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: The pre-training and fine-tuning of the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

[0065] In one embodiment, when the processor executes the computer program, the processor further implements the following steps: The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

[0066] For the specific limitations on the steps implemented when the processor executes the computer program, please refer to the limitations on the method for tracing the source of welding problems mentioned above, which will not be repeated here.

[0067] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

[0068] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; the quality manufacturing data includes the quality manufacturing data extracted from the hull section manufacturing quality database and the defective product quality database.

[0069] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

[0070] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The selecting process parameters according to the various welding process data sets to construct a welding problem tracing model includes: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

[0071] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

[0072] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The pre-training and fine-tuning of the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

[0073] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented: The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

[0074] For the specific limitations on the steps implemented when the computer program is executed by the processor, please refer to the limitations on the method for tracing the source of welding problems mentioned above, which will not be repeated here.

[0075] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0076] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0077] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A welding problem tracing method, characterized in that: include: Collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and roughly classify the data according to the relevance between the data and the process quality to form roughly classified data; Performing data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and storing them in a database; Selecting process parameters according to the various welding process data sets to build a welding problem traceability model, and training the welding problem traceability model based on the welding process quality traceability process; The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of welding defective products are traced and output.

2. The welding problem tracing method according to claim 1 is characterized in that: The terminal industrial computer process data includes flat butt welding process data and corner butt welding process data; the business information system data is the quality data of the product throughout its life cycle; the quality manufacturing data includes the quality manufacturing data extracted from the hull section manufacturing quality database and the defective product quality database.

3. The welding problem tracing method according to claim 1 is characterized in that: The data cleaning and data standardization of the coarse classification data includes: Deleting duplicate information, correcting errors, and maintaining data consistency for missing values, noise data, inconsistent data, and test data in the coarse classification data; The cleaned data is assigned default values, format / type / length conversion, code / value / specific character conversion.

4. The welding problem tracing method according to claim 1 is characterized in that: The selecting process parameters according to the various welding process data sets to construct a welding problem tracing model comprises: Tracing the process parameters in the various welding process data sets to obtain process parameters that affect welding quality, the process parameters including welding current, welding voltage, laser power, welding speed and wire feeding speed; A welding problem tracing model structure is constructed using Deep Belief Networks (DBNs), wherein the Deep Belief Networks are stacked by multiple Restricted Boltzmann Machines (RBMs), and the Restricted Boltzmann Machines are composed of two layers of fully connected neurons; the welding problem tracing model includes an input layer, multiple hidden layers and an output layer; The weights of the neurons of the welding problem tracing model are set according to the weights of the process parameters, the neuron weights are initialized, and an activation function is selected.

5. The welding problem tracing method according to claim 4 is characterized in that: The welding problem tracing model training based on the welding process quality tracing process includes: The welding process quality traceability process is set to perform flaw detection on qualified samples, determine the properties of the flaw detection results, record the process parameters if it is determined to be a qualified product, and determine the influencing factors of the welding problem traceability model if it is determined to be an unqualified product, adjust the process parameters and retest the qualified samples for flaw detection; According to the welding process quality traceability process, forward propagation, backward propagation, gradient calculation and weight update are performed on the welding problem traceability model, wherein the forward propagation is the activation from the output layer to the hidden layer, the backward propagation is the reconstruction from the hidden layer to the output layer, the gradient calculation is the gradient of weight update calculated by contrast divergence, and the weight update is the weight update by learning rate; The welding problem tracing model is pre-trained and fine-tuned according to the training data set; the pre-training includes greedily training each restricted Boltzmann machine layer layer by layer in order from bottom to top, using the contrast divergence algorithm to train the restricted Boltzmann machine layer, and after each restricted Boltzmann machine layer is trained, its weight is used as the input of the next restricted Boltzmann machine layer; the fine-tuning includes using the back propagation algorithm for supervised learning, and minimizing the prediction error of the training data by adjusting the weight.

6. The welding problem tracing method according to claim 1 is characterized in that: The pre-training and fine-tuning of the welding problem tracing model according to the training data set includes: Initializing the weight values ​​of the process parameters of the welding problem tracing model, inputting the training data set into the welding problem tracing model, calculating the result, and comparing it with the expected result to determine whether the result is satisfactory; If the result is satisfactory, the welding problem tracing model is tested and then applied; If the result is not satisfactory, the weight value of the process parameter is adjusted by gradient descent, and the training data set is re-input into the welding problem tracing model to calculate the result.

7. The welding problem tracing method according to claim 1 is characterized in that: The trained welding problem tracing model is used to locate and sort the influencing factors, and the process parameters and the proportion of influencing factors of the welding defective products are obtained by tracing and outputting the following: Based on the normal distribution, the 3sigma criterion is used to obtain the data exceeding 3sigma in the various welding process data sets as abnormal point data. If the abnormal point data is between the mean plus or minus three times the standard deviation, it is considered to be normal data, otherwise, it is considered to be abnormal data; The influencing factors are determined according to the abnormal data, and the influencing factors are sorted, and the process parameters and the proportion of influencing factors of the welding defective products are traced and output.

8. A welding problem tracing device, characterized in that: The device comprises: The data rough classification module is used to collect the process data of the terminal industrial computer, the business information system data, and the quality manufacturing data, and to roughly classify the data according to the correlation between the data and the process quality to form rough classification data; A welding process data set processing module is used to perform data cleaning and data standardization on the coarse classification data to obtain various welding process data sets and store them in a database; A model building and training module, used to select process parameters according to the various welding process data sets to build a welding problem tracing model, and to train the welding problem tracing model based on the welding process quality tracing process; The model application module is used to use the trained welding problem tracing model to locate and sort the influencing factors, trace the process parameters of welding defective products and the proportion of their influencing factors, and output them.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images