A method for evaluating steel structure welding process quality based on big data processing
The method addresses data preprocessing and fusion challenges in welding by using deep learning to predict and optimize welding quality, ensuring consistent product quality and reducing waste and costs.
Patent Information
- Application Number
- CN202411348024.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-26
- Publication Date
- 2024-12-27
- Estimated Expiration
- 2044-09-26
AI Technical Summary
Existing welding technologies lack effective data preprocessing to account for environmental dynamics, outlier detection, and multimodal data fusion, leading to data noise and limited model adaptability in complex welding scenarios, resulting in inconsistent quality control.
A method involving data collection, preprocessing, and multimodal data fusion using deep learning models to extract welding quality features, construct a prediction model, and optimize environmental parameters and materials based on quality evaluation results, stored in a database for continuous improvement.
Enhances welding quality prediction accuracy, ensures product consistency, reduces scrap rates and costs, and optimizes the welding process through real-time monitoring and data analysis.
Smart Images
Figure CN118875566B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent welding, and in particular to a method for evaluating the quality of steel structure welding process based on big data processing. Background Art
[0002] With the continuous advancement of industrial automation and informatization, the accuracy and efficiency of welding processes have become important indicators for measuring product quality. The quality control of traditional welding processes mainly relies on manual inspection and experience judgment. This method has problems such as strong subjectivity, low efficiency, and poor consistency. In recent years, with the rise of the Internet of Things (IoT), big data analysis, and artificial intelligence (AI) technologies, the quality control of welding processes has begun to shift towards intelligence. The advancement of sensor technology has made it possible to monitor the welding environment in real time, and the application of big data processing and deep learning algorithms can effectively explore the potential laws in welding data and achieve accurate prediction and optimization of welding quality.
[0003] Although sensor-based real-time monitoring and big data analysis have shown great potential in improving welding quality, existing technologies still have several limitations. First, the data preprocessing link often ignores the dynamic impact of environmental factors on welding quality, lacks effective outlier detection and data smoothing strategies, resulting in data noise interfering with model training. Second, single environmental data or welding parameters are difficult to fully reflect welding quality, and multimodal data fusion technology is not yet mature, which limits the model's ability to adapt to complex welding scenarios. Summary of the invention
[0004] In view of the above existing problems, the present invention is proposed.
[0005] Therefore, the present invention provides a steel structure welding process quality evaluation method based on big data processing to solve the problems of shortcomings of existing welding technology in data processing, model effectiveness and process optimization.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0007] In the first aspect, an embodiment of the present invention provides a method for evaluating the quality of steel structure welding process based on big data processing, which includes collecting environmental data and preprocessing it; inputting the preprocessed environmental data into a deep learning model to extract welding quality features; performing multimodal data fusion on the welding quality features and the preprocessed environmental data to form a comprehensive feature vector; constructing a welding quality prediction model based on the comprehensive feature vector to predict and evaluate the welding quality level; optimizing environmental parameters and welding materials based on the welding quality prediction value and the quality evaluation result; storing and analyzing the optimized environmental parameters and welding quality level in a database to generate a welding process quality report.
[0008] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, the specific steps of collecting environmental data and preprocessing are as follows:
[0009] Collect environmental data in the welding area through various types of sensors, including rapidly changing environmental conditions during welding;
[0010] The standardized value of each measurement point is calculated by the Z-Score method, and the environmental data are smoothed according to the standardized value.
[0011] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, wherein: the pre-processed environmental data is input into the deep learning model to extract welding quality features, and the specific steps are as follows:
[0012] The preprocessed environmental data are combined into a time series feature matrix F, where each column represents all sensor readings at a time point, and each row represents the readings of a sensor at all time points;
[0013] LSTM is selected as the deep learning model to process sequence data from both forward and reverse directions, capture the forward and backward dependencies in the time series data, and define a function Φ(F; Θ) to extract welding quality features from the time series feature matrix F. The expression is:
[0014]
[0015] in, is the hidden state of the forward LSTM at the last time point T, is the hidden state of the reverse LSTM at the last time point T, W q is the weight matrix, b q is the bias vector, Q is the welding quality feature vector, Θ is the parameter of the deep learning model, and ct represents the splicing operation.
[0016] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, wherein: the welding quality characteristics and the pre-processed environmental data are multi-modal data fused to form a comprehensive feature vector, and the specific steps are as follows:
[0017] By copying the welding quality feature vector Q so that each column of it corresponds to each column of the time series matrix F, a new feature matrix Q′ is formed;
[0018] A polymorphic fusion function G(Q′, F; Ω) is designed to fuse the information of the feature matrix Q′ and the time series matrix F to generate a comprehensive feature matrix C, which is expressed as:
[0019]
[0020] Among them, Ω is the parameter of the fusion function, d is the feature dimension, sx is a nonlinear function, is the matrix transpose operator;
[0021] By averaging each column in the comprehensive feature matrix C, the comprehensive feature vector V is extracted, and the expression is:
[0022]
[0023] Where D is the number of columns of the feature matrix C, C i1 is the element in the i-th row and the first column of the comprehensive feature matrix C. i2 is the element in the i-th row and the second column of the comprehensive feature matrix C, Z is the length of the time series, C iD It is the element in the i-th row and D-th column of the comprehensive feature matrix C.
[0024] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, a welding quality prediction model is constructed based on a comprehensive feature vector to predict the welding quality level. The specific steps are as follows:
[0025] Since welding quality presents complex nonlinear variation characteristics over time, that is, as time goes by, welding quality may not follow a simple linear law, but will be affected by a combination of factors, which may interact in a nonlinear way. Therefore, a nonlinear activation function f that can handle the nonlinear variation of welding quality over time is used to design a welding quality prediction model.
[0026] The complex patterns contained in the comprehensive feature vector V are analyzed by the welding quality prediction model and converted into the prediction of the welding quality level Y;
[0027] In the training phase, the welding quality prediction model receives the comprehensive feature vector V as input and predicts the welding quality level Y corresponding to each vector. The training process involves the adjustment of the weight matrix W and the bias term b, and the calculation minimizes the difference between the predicted value and the actual welding quality.
[0028] After training, the welding quality prediction model can receive the new comprehensive feature vector V and output the predicted welding quality level Y, which is expressed as:
[0029] Y = f(W·V+b);
[0030] Among them, f is a non-linear activation function.
[0031] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, the welding quality level is evaluated according to the prediction results, and the specific steps are as follows:
[0032] Based on previous welding quality standards and historical data, a welding quality standard T is defined, the welding quality prediction Y and the welding quality standard T are analyzed to obtain analysis results, and the welding quality level is evaluated based on the analysis results;
[0033] If Y ≥ T, it indicates that the welding quality level has reached the predetermined standard and can continue to be used for further assembly and use;
[0034] If Y < T, it indicates that the welding quality level does not meet the predetermined standard, and the welded parts will be marked for review, requiring further optimization of environmental parameters and welding materials;
[0035] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, wherein: based on the welding quality prediction value and the quality evaluation result, the environmental parameters and welding materials are optimized, and the specific steps are as follows:
[0036] In the prediction result Y output by the welding quality prediction model, the specific value of each prediction is recorded, as well as whether it exceeds the set welding quality standard T;
[0037] Summarize the collected prediction results, calculate the ratio of qualified to unqualified, and the changing trend of welding quality under different environments;
[0038] The cost analysis of the welding process is as follows:
[0039]
[0040] Among them, w l is the weight of the lth cost element, h li is the actual cost value of the lth cost element, N is the total number of cost elements, and H is the cost;
[0041] The resource utilization analysis of the welding process is expressed as follows:
[0042]
[0043] Among them, r j is the utilization rate of the jth resource, u j is the weight of the jth resource, M is the total number of resource types, and R is the resource utilization score;
[0044] An optimization objective function U is set to optimize environmental parameters and welding materials based on the analysis of the prediction results Y, the cost score H and the resource utilization score R. The expression is:
[0045] U=w 1 ·Y+w 2 ·H+w 3 ·R;
[0046] Among them, Y represents the average score of welding quality, H represents the cost score, R represents the resource utilization score, and w 1 is the weight of E, w 2 is the weight of H, w 3 is the weight of R;
[0047] Environmental parameters and welding material properties are represented as a variable set P = [P 1 , P 2 , ..., P n ], where P n Represents the nth element in the variable set P;
[0048] Using the optimization algorithm, we can find the parameter set that can maximize the optimization objective function U. The expression is:
[0049]
[0050] Among them, P * The optimized parameter set;
[0051] The optimal parameter set P * Applied to the actual welding process, adjust environmental parameters and select corresponding welding materials.
[0052] As a preferred solution of the steel structure welding process quality evaluation method based on big data processing described in the present invention, wherein: the optimized environmental parameters and welding quality level are stored in a database and analyzed to generate a welding process quality report, the specific steps are as follows:
[0053] Collect key data points from optimized environmental parameters and weld quality levels;
[0054] A new data record is created for each welding process, and all optimized parameters are combined into a single data record. * , the predicted welding quality level Y, and the evaluation result obtained by comparing Y with the welding quality standard T;
[0055] Add the created data records to the database, retrieve the necessary data related to the welding process quality from the database, conduct in-depth analysis on the retrieved necessary data, identify trends, anomalies and improvement points in the welding process, and write a detailed welding process quality report based on the results of the data analysis.
[0056] In a second aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the steel structure welding process quality evaluation method based on big data processing as described in the first aspect of the present invention is implemented.
[0057] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the steel structure welding process quality evaluation method based on big data processing as described in the first aspect of the present invention is implemented.
[0058] The beneficial effects of the present invention are as follows: through the deep learning model, the key features of welding quality are extracted from time series data to achieve a deep understanding and prediction of the welding process; through the multimodal data fusion technology, the environmental data is combined with the welding quality features to enhance the decision-making ability of the prediction model; based on the comprehensive feature vector, the constructed welding quality prediction model accurately predicts the welding quality level; by evaluating the welding quality, it is ensured that the product quality meets the standards; by collecting the welding quality prediction and quality evaluation results, the environmental parameters and welding material selection are optimized, and the continuous improvement of the welding process is achieved; all data and analysis results are stored in the database, which is convenient for in-depth analysis of the trends, anomalies and improvement points of the welding process, thereby continuously optimizing the welding process, improving production efficiency and product quality, reducing production costs and resource waste, ensuring the safety and reliability of the welding structure, and improving the intelligence level of the welding process. BRIEF DESCRIPTION OF THE DRAWINGS
[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0060] Figure 1 This is a flow chart of the steel structure welding process quality evaluation method based on big data processing in Example 1.
[0061] Figure 2 This is a flow chart for generating a welding process quality report in Example 1. DETAILED DESCRIPTION
[0062] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.
[0063] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0064] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0065] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and this embodiment provides a steel structure welding process quality evaluation method based on big data processing, comprising the following steps: collecting environmental data and preprocessing it; inputting the preprocessed environmental data into a deep learning model to extract welding quality features; performing multimodal data fusion on the welding quality features and the preprocessed environmental data to form a comprehensive feature vector; constructing a welding quality prediction model based on the comprehensive feature vector to predict and evaluate the welding quality level; optimizing environmental parameters and welding materials based on the welding quality prediction values and quality evaluation results; storing the collected welding quality prediction values and quality evaluation results in a database, and generating a welding process quality report based on the information in the database.
[0066] This embodiment also provides a steel structure welding process quality evaluation system based on big data processing, including:
[0067] S1. Collect environmental data and perform preprocessing.
[0068] Furthermore, multiple types of sensors are installed in the welding area, including temperature sensors, humidity sensors, wind speed sensors, pressure sensors, and gas concentration detectors (such as oxygen and carbon dioxide); the sampling frequency of the sensors is set to capture the rapidly changing environmental conditions during the welding process;
[0069] Use LoRaWAN to transmit data to a central server to reduce latency and data loss. Apply the Z-Score method to calculate the normalized value of each measurement point and remove outliers. Then use linear interpolation to fill in the removed data points to maintain data continuity. Use a moving average filter to smooth the data and reduce the impact of noise.
[0070] The preprocessed environmental data is divided into training set, validation set and test set for model training, parameter adjustment and performance evaluation.
[0071] It should also be noted that the Z-Score expression is:
[0072]
[0073] Where X is the measured value, μ is the mean value, and σ is the standard deviation;
[0074]
[0075] Among them, S t is the smoothed data point, x i is the original data point, n is the window size, and t represents the position or time index of the current data point.
[0076] S2. Input the preprocessed environmental data into the deep learning model to extract welding quality features.
[0077] Furthermore, the preprocessed environmental data are combined into a time series feature matrix F where each column represents all sensor readings at a time point, and each row represents the readings of a sensor at all time points;
[0078] LSTM is selected as the deep learning model to process sequence data from both forward and reverse directions, capture the forward and backward dependencies in time series data, and understand the sequence characteristics more comprehensively. A function Φ(F; Θ) is defined to extract welding quality features from the time series feature matrix F. The expression is:
[0079]
[0080] in, is the hidden state of the forward LSTM at the last time point T, is the hidden state of the reverse LSTM at the last time point T, W q is the weight matrix, b q is the bias vector, Q is the welding quality feature vector, Θ is the parameter of the deep learning model, and ct represents the splicing operation;
[0081] It should also be noted that the range of the eigenvector Q depends on the weight matrix W q and the bias vector b q After the training is completed, each element of Q represents a different aspect of welding quality, such as welding strength, welding stability, the possibility of welding defects, etc. The value reflects the strength of these aspects. It is usually normalized and the value range is between [-1, 1]. The larger the absolute value, the more obvious the feature is or the more serious the problem is.
[0082] S3. Perform multimodal data fusion on the welding quality features and the pre-processed environmental data to form a comprehensive feature vector.
[0083] Furthermore, by replicating the welding quality feature vector Q so that each column of it corresponds to each column of the time series matrix F, a new feature matrix Q′ is formed, in which each column of Q is a replica of Q;
[0084] A polymorphic fusion function G(Q′, F; Ω) is designed to fuse the information of the feature matrix Q′ and the time series matrix F to generate a comprehensive feature matrix C, which is expressed as:
[0085]
[0086] Among them, Ω is the parameter of the fusion function, d is the feature dimension, sx is a nonlinear function, is the matrix transpose operator;
[0087] By averaging each column in the comprehensive feature matrix C, the comprehensive feature vector V is extracted, and the expression is:
[0088]
[0089] Where D is the number of columns of the feature matrix C, C i1 is the element in the i-th row and the first column of the comprehensive feature matrix C. i2 is the element in the i-th row and the second column of the comprehensive feature matrix C, Z is the length of the time series, C iD is the element in the i-th row and D-th column of the comprehensive feature matrix C;
[0090] It should also be noted that the dot product of Q′ and F The similarity matrix between Q′ and F is calculated. Each element represents the similarity between a column of Q′ (i.e., a copy of the welding quality feature vector) and a row of F (i.e., the environmental data feature). The division is used to avoid excessive values and ensure numerical stability.
[0091] The range of the comprehensive feature matrix C depends on the initial range of F and the weight distribution of Q′. In this example, since the range of Q′ has been normalized to [-1, 1] and the range of F may be wider, the range of V may be outside [-1, 1]. However, through appropriate normalization processing (such as maximum and minimum scaling), the range of V can be rescaled to [-1, 1] or other required ranges. Each value represents the correlation and importance of a certain feature in the welding environment to the welding quality feature.
[0092] S4. Construct a welding quality prediction model based on the comprehensive feature vector to predict the welding quality level and conduct an evaluation.
[0093] Furthermore, a nonlinear activation function f that can handle the nonlinear change of welding quality over time is used to design a welding quality prediction model;
[0094] The complex patterns contained in the comprehensive feature vector V are analyzed by the welding quality prediction model and converted into the prediction of the welding quality level Y;
[0095] In the training phase, the welding quality prediction model receives the comprehensive feature vector V as input and predicts the welding quality level Y corresponding to each vector. The training process involves the adjustment of the weight matrix W and the bias term b, and the calculation minimizes the difference between the predicted value and the actual welding quality.
[0096] After training, the welding quality prediction model can receive the new comprehensive feature vector V and output the predicted welding quality level Y, which is expressed as:
[0097] Y = f(W·V+b);
[0098] Among them, f is a nonlinear activation function, which is used to introduce nonlinear characteristics and increase the expressiveness of the model;
[0099] Based on previous welding quality standards and historical data, a welding quality standard T is defined, the welding quality prediction Y and the welding quality standard T are analyzed to obtain analysis results, and the welding quality level is evaluated based on the analysis results;
[0100] If Y ≥ T, it indicates that the welding quality level has reached the predetermined standard and can continue to be used for further assembly and use;
[0101] If Y < T, it indicates that the welding quality level does not meet the predetermined standard, and the welded parts will be marked for review, requiring further optimization of environmental parameters and welding materials;
[0102] It should also be noted that this threshold can be determined empirically or obtained by analyzing the distribution of welding quality data. For example, if the value range of Y is standardized to [-1,1], then T=0 can be set as the reference point. A positive number indicates that the quality is higher than the average level, and a negative number indicates that the quality is lower than the average level.
[0103] Evaluate the welding quality level based on the prediction results. For example, if Y is a scalar, you can set a threshold to determine whether the welding quality is qualified. If it is a vector, you can analyze the value of each dimension to understand the possibility and degree of specific defects.
[0104] For each prediction result, evaluation metrics such as precision, recall, F1 score, or root mean square error are calculated to quantify the prediction performance of the welding quality prediction model.
[0105] S5. Optimize environmental parameters and welding materials based on welding quality prediction values and quality assessment results.
[0106] Furthermore, in the prediction result Y output by the welding quality prediction model, the specific value of each prediction is recorded, as well as whether it exceeds the set welding quality standard T;
[0107] Summarize the collected prediction results, calculate the ratio of qualified to unqualified, and the changing trend of welding quality under different environments;
[0108] The cost analysis of the welding process is as follows:
[0109]
[0110] Among them, w l is the weight of the lth cost element, h l is the actual cost value of the lth cost element, N is the total number of cost elements, and H is the cost;
[0111] The resource utilization analysis of the welding process is expressed as follows:
[0112]
[0113] Among them, r j is the utilization rate of the jth resource, u j is the weight of the jth resource, M is the total number of resource types, and R is the resource utilization score;
[0114] An optimization objective function U is set to optimize environmental parameters and welding materials based on the analysis of the prediction results Y, the cost score H and the resource utilization score R. The expression is:
[0115] U=w 1 ·Y+w 2 ·H+w 3 ·R;
[0116] Among them, Y represents the average score of welding quality, H represents the cost score, R represents the resource utilization score, and w 1 is the weight of E, w 2 is the weight of H, w 3 is the weight of R;
[0117] The environmental parameters and welding material properties are represented as a variable set P = P = [P 1 , P 2 , ..., P n ], where P n Represents the nth element in the variable set P;
[0118] Using the optimization algorithm, we can find the parameter set that can maximize the optimization objective function U. The expression is:
[0119]
[0120] Among them, P * The optimized parameter set;
[0121] The optimal parameter set P * Applied to the actual welding process, adjust environmental parameters and select corresponding welding materials.
[0122] It should also be noted that in the welding quality prediction, the prediction result Y output by the model needs to be recorded and judged whether it exceeds the set standard T, the prediction results are summarized, and the pass rate and environmental impact trend are statistically analyzed. Define the weighted sum of the optimization objective function U, the comprehensive welding quality E, the cost H and the resource utilization rate R, and the weight w 1 , w 2 , w 3 , reflecting the importance of each indicator. Environmental parameters and material properties constitute the variable set P, and the optimization algorithm is used to solve the new parameter set P that maximizes U * Then P * Applied to the welding process, it dynamically adjusts the environment and material selection to achieve high-quality welding while optimizing costs and resources.
[0123] S6. The optimized environmental parameters and welding quality levels are stored in the database and analyzed to generate a welding process quality report.
[0124] Collect key data points from optimized environmental parameters and weld quality levels;
[0125] A new data record is created for each welding process, and all optimized parameters are combined into a single data record. * (such as temperature, humidity, wind speed, pressure, gas concentration, alloy composition, thickness, preheating temperature, etc.), the predicted welding quality level Y, and the evaluation result compared with the welding quality standard T;
[0126] Add the created data records to the database and retrieve the necessary data related to the welding process quality from the database, including all welding records over a period of time or welding records under specific conditions. Perform in-depth analysis on the retrieved data to identify trends, anomalies and improvement points in the welding process. For example, you can analyze which environmental parameter changes have the greatest impact on welding quality, or which operator has the highest welding quality. Based on the results of the data analysis, write a detailed welding process quality report.
[0127] This embodiment also provides a computer device, which is suitable for the steel structure welding process quality evaluation method based on big data processing, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the steel structure welding process quality evaluation method based on big data processing proposed in the above embodiment.
[0128] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. 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 and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.
[0129] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for evaluating the quality of welding process of steel structures based on big data processing as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.
[0130] In summary, the present invention extracts key features of welding quality from time series data through a deep learning model, achieves a deep understanding and prediction of the welding process, combines environmental data with welding quality features through multimodal data fusion technology, enhances the decision-making ability of the prediction model, and constructs a welding quality prediction model based on a comprehensive feature vector to accurately predict the welding quality level. By evaluating the welding quality, it ensures that the product quality meets the standards. By collecting welding quality prediction and quality assessment results, it optimizes environmental parameters and welding material selection, and achieves continuous improvement of the welding process. All data and analysis results are stored in a database, which is convenient for in-depth analysis of trends, anomalies and improvement points of the welding process, thereby continuously optimizing the welding process, improving production efficiency and product quality, reducing production costs and resource waste, ensuring the safety and reliability of the welding structure, and improving the intelligence level of the welding process.
[0131] Example 2, referring to Table 1, is the second example of the present invention. In order to further verify the technical solution of the present invention, experimental simulation data of a steel structure welding process quality evaluation method based on big data processing is provided.
[0132] Four welding samples were selected and tested under four different environmental conditions (temperature, humidity, wind speed, pressure, oxygen concentration, CO 2 The welding was carried out under the conditions of 100 nm and 80 nm concentration, and the welding quality prediction method of the present invention was used to predict the welding quality of the five welding samples.
[0133] Firstly, sensors are installed to collect environmental parameters during welding, missing data points are filled using linear interpolation technology, and the data are smoothed using a moving average filter to further reduce noise and improve data quality.
[0134] Secondly, the LSTM deep learning model is used to process the preprocessed environmental data and extract the welding quality features. The value range of the feature vector Q is between [-1,1]. Each element reflects different aspects of the welding quality. The welding quality features are combined with the environmental data to form a comprehensive feature vector V.
[0135] Next, a welding quality prediction model is constructed, and the comprehensive feature vector V is processed using a nonlinear activation function to predict the welding quality level Y. The model is trained by minimizing the difference between the predicted value and the actual welding quality.
[0136] Finally, the prediction results and quality assessment are collected, and the environmental parameters and welding materials are optimized based on the results. When the prediction result Y meets the welding quality standard T, the welded part is considered qualified; otherwise, a detailed inspection and possible repair are required.
[0137] The details are shown in Table 1:
[0138] Table 1 Experimental record table
[0139]
[0140] Through the data analysis of the above table, it can be seen that the new invention method is significantly better than the traditional method in predicting the welding quality level. In all five welding cases, the Ynew value predicted by the new method is higher than the Ytraditional value predicted by the traditional method. For example, under the conditions of welding 1 and welding 3, the predicted values of the new method are improved by 0.4 and 0.2 respectively, which shows that the new method can more accurately identify favorable conditions in the welding process, thereby improving the prediction accuracy of welding quality.
[0141] The steel structure welding process quality evaluation method of the present invention can not only realize real-time monitoring and data analysis of key parameters in the production process to ensure product consistency and stability, but also predict potential welding process deviations and quality problems, and take corrective measures in advance, thereby significantly reducing scrap rates and production costs. This reflects the innovation and practicality of the present invention, which not only promotes the transformation of traditional manufacturing to intelligent manufacturing, but also promotes the integration of interdisciplinary knowledge and technology, laying a solid foundation for the future Industry 4.0 and smart factories.
[0142] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A steel structure welding process quality evaluation method based on big data processing, characterized by: include, Collect environmental data and pre-process it; The pre-processed environmental data is fed into the deep learning model to extract welding quality features; Multimodal data fusion is performed on welding quality features and pre-processed environmental data to form a comprehensive feature vector; Construct a welding quality prediction model based on the comprehensive feature vector to predict and evaluate the welding quality level; Optimize environmental parameters and welding materials based on welding quality prediction values and quality assessment results; The optimized environmental parameters and welding quality levels are stored in the database and analyzed to generate a welding process quality report; The pre-processed environmental data is input into the deep learning model to extract welding quality features. The specific steps are as follows: Combine the preprocessed environmental data into a time series feature matrix F, where each column represents all sensor readings at a time point, and each row represents the readings of a sensor at all time points; LSTM is selected as the deep learning model to process sequence data from both forward and reverse directions, capture the forward and backward dependencies in the time series data, and define a function Φ(F; Θ) to extract welding quality features from the time series feature matrix F. The expression is: in, is the hidden state of the forward LSTM at the last time point T, is the hidden state of the reverse LSTM at the last time point T, W q is the weight matrix, b q is the bias vector, Q is the welding quality feature vector, Θ is the parameter of the deep learning model, and ct represents the splicing operation.
2. The method for evaluating the quality of steel structure welding process based on big data processing according to claim 1, characterized in that: The specific steps of collecting environmental data and preprocessing are as follows: Collect environmental data in the welding area through various types of sensors, including rapidly changing environmental conditions during welding; The standardized value of each measurement point is calculated by the Z-Score method, and the environmental data are smoothed according to the standardized value.
3. The method for evaluating the quality of steel structure welding process based on big data processing according to claim 2, characterized in that: The welding quality features are fused with the pre-processed environmental data to form a comprehensive feature vector. The specific steps are as follows: By copying the welding quality feature vector Q so that each column of it corresponds to each column of the time series matrix F, a new feature matrix Q′ is formed; A polymorphic fusion function G(Q′, F; Ω) is designed to fuse the information of the feature matrix Q′ and the time series matrix F to generate a comprehensive feature matrix C, which is expressed as: Among them, Ω is the parameter of the fusion function, d is the feature dimension, sx is a nonlinear function, is the matrix transpose operator; By averaging each column in the comprehensive feature matrix C, the comprehensive feature vector V is extracted, and the expression is: Where D is the number of columns of the feature matrix C, C i1 is the element in the i-th row and the first column of the comprehensive feature matrix C. i2 is the element in the i-th row and the second column of the comprehensive feature matrix C, Z is the length of the time series, C iD It is the element in the i-th row and D-th column of the comprehensive feature matrix C.
4. The method for evaluating the quality of steel structure welding process based on big data processing according to claim 3, characterized in that: Based on the comprehensive feature vector, a welding quality prediction model is constructed to predict the welding quality level. The specific steps are as follows: A nonlinear activation function f that can handle the nonlinear change of welding quality over time is used to design a welding quality prediction model. The complex patterns contained in the comprehensive feature vector V are analyzed by the welding quality prediction model and converted into the prediction of the welding quality level Y; In the training phase, the welding quality prediction model receives the comprehensive feature vector V as input and predicts the welding quality level Y corresponding to each vector. The training process involves the adjustment of the weight matrix W and the bias term b, and the calculation minimizes the difference between the predicted value and the actual welding quality. After training, the welding quality prediction model receives the new comprehensive feature vector V and outputs the predicted welding quality level Y, which is expressed as: Y = f(W·V+b); Among them, f is a non-linear activation function.
5. The method for evaluating the quality of steel structure welding process based on big data processing according to claim 4, characterized in that: Evaluate the welding quality level based on the prediction results. The specific steps are as follows: Based on previous welding quality standards and historical data, a welding quality standard T is defined, the welding quality prediction Y and the welding quality standard T are analyzed to obtain analysis results, and the welding quality level is evaluated based on the analysis results; If Y ≥ T, it indicates that the welding quality level has reached the predetermined standard and can continue to be used for further assembly and use; If Y<T, it indicates that the welding quality level does not meet the predetermined standard, and the welded parts will be marked for review, requiring further optimization of environmental parameters and welding materials.
6. The method for evaluating the quality of steel structure welding process based on big data processing according to claim 5, characterized in that: The specific steps of optimizing environmental parameters and welding materials based on the welding quality prediction value and quality assessment result are as follows: In the prediction result Y output by the welding quality prediction model, the specific value of each prediction is recorded, as well as whether it exceeds the set welding quality standard T; Summarize the collected prediction results, calculate the ratio of qualified to unqualified, and the changing trend of welding quality under different environments; The cost analysis of the welding process is as follows: Among them, W l is the weight of the lth cost element, h l is the actual cost value of the lth cost element, N is the total number of cost elements, and H is the cost; The resource utilization analysis of the welding process is expressed as follows: Among them, r j is the utilization rate of the jth resource, u j is the weight of the jth resource, M is the total number of resource types, and R is the resource utilization score; An optimization objective function U is set to optimize environmental parameters and welding materials based on the analysis of the prediction results Y, the cost score H and the resource utilization score R. The expression is: U=w1·Y+w2·H+w3·R; Among them, Y represents the average score of welding quality, H represents the cost score, R represents the resource utilization score, w1 is the weight of E, w2 is the weight of H, and w3 is the weight of R; The environmental parameters and welding material properties are represented as a variable set P = [P1, P2, ..., P n ], where P n Represents the nth element in the variable set P; Using the optimization algorithm, we can find the parameter set that can maximize the optimization objective function U. The expression is: Among them, P * The optimized parameter set; The optimal parameter set P * Applied to the actual welding process, adjust environmental parameters and select corresponding welding materials.
7. The method for evaluating the quality of steel structure welding process based on big data processing according to claim 6, characterized in that: The optimized environmental parameters and welding quality level are stored in the database and analyzed to generate a welding process quality report. The specific steps are as follows: Collect key data points from optimized environmental parameters and weld quality levels; For each welding process, a new data record is created, including the optimized parameter set P*, the predicted welding quality level Y, and the evaluation result of Y compared with the welding quality standard T; Add the created data records to the database, retrieve the necessary data related to the welding process quality from the database, conduct in-depth analysis on the retrieved necessary data, identify trends, anomalies and improvement points in the welding process, and write a detailed welding process quality report based on the results of the data analysis.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the steel structure welding process quality evaluation method based on big data processing according to any one of claims 1 to 7 are implemented.
9. 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 steel structure welding process quality evaluation method based on big data processing according to any one of claims 1 to 7 are implemented.