Welding fume intelligent monitoring system and method based on Internet of Things

By installing sensors in the welding workshop and using IoT and machine learning technology to monitor welding smoke concentration in real time, the problem of the inability to monitor the impact of environmental conditions in real time is solved, and the health and production safety of welders are improved.

CN118898036AInactive Publication Date: 2024-11-05JIANGSU PENGCHENG IND INTELLIGENT TECH CO LTD
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202410951250.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-16
Publication Date
2024-11-05
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing welding smoke monitoring methods cannot achieve real-time monitoring and are sensitive to environmental conditions, making it difficult to accurately reflect the smoke concentration in the welding workshop, resulting in threats to welders' health and increased risk of occupational diseases.

Method used

Using an intelligent monitoring system based on the Internet of Things, by installing dust concentration detectors, temperature and humidity sensors and wind speed sensors in the welding workshop, data from multiple time points is collected, and machine learning technology is used to extract and analyze the data to generate classification results. To determine whether the smoke concentration exceeds the safety standards.

Benefits of technology

Real-time monitoring and early warning of smoke and dust in welding workshops is achieved, reducing occupational disease risks and improving the safety and sustainability of the production environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118898036A_ABST
    Figure CN118898036A_ABST
Patent Text Reader

Abstract

The invention relates to the field of intelligent monitoring of welding fume, and particularly discloses an intelligent welding fume monitoring system and method based on the Internet of Things, and the method comprises the steps: firstly, installing a dust concentration detector, a temperature and humidity sensor and a wind speed sensor at each monitoring position in a welding workshop; collecting smoke dust concentration values at a plurality of preset time points of each monitoring position in the welding workshop, environment data at a plurality of preset time points of each monitoring position in the welding workshop and safety standard information about exposure of harmful substances in the welding workshop; and feature extraction and analysis are carried out on the input data by using a machine learning technology to obtain a classification result, wherein the classification result is used for representing whether the welding fume concentration in the welding workshop exceeds a safety standard or not. Therefore, according to the classification result, real-time monitoring and early warning of the smoke dust in the welding workshop are facilitated, so that the health of welders is guaranteed, the occupational disease risk is reduced, and the safety and sustainability of the production environment are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent monitoring of welding fumes, and more specifically, to an intelligent monitoring system and method for welding fumes based on the Internet of Things. Background Art

[0002] In welding workshops, a large amount of dust and smoke is generated, mainly from welding processes such as spot welding, CO2 gas shielded welding and argon arc welding. In particular, CO2 gas shielded welding and argon arc welding will release a large amount of welding smoke and arc light. In other words, in welding production, a large amount of harmful metal smoke, electric welding dust and toxic gases will be released. When the concentration of these substances exceeds the safety standard, it may cause harm to the health of welders and lead to occupational diseases such as pneumoconiosis, manganese poisoning and metal fever.

[0003] At present, the methods for monitoring welding smoke include chemical analysis, optical methods, etc. Chemical analysis usually requires sample collection for laboratory testing or analysis, and cannot be monitored in real time; while optical methods are sensitive to environmental conditions, such as light intensity, temperature, humidity and other factors that may affect the accuracy of the monitoring results.

[0004] Therefore, we look forward to an intelligent welding fume monitoring system and method based on the Internet of Things, which uses machine learning technology to analyze the fume situation in the welding workshop, so as to achieve real-time monitoring and early warning of the fume in the welding workshop, thereby protecting the health of welders, reducing the risk of occupational diseases, and improving the safety and sustainability of the production environment. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent monitoring system and method for welding smoke based on the Internet of Things. First, a dust concentration detector, a temperature and humidity sensor, and a wind speed sensor are installed at each monitoring position in the welding workshop to collect smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and safety standard information on exposure to harmful substances in the welding workshop; then, machine learning technology is used to extract and analyze these input data to obtain classification results, and the classification results are used to indicate whether the welding smoke concentration in the welding workshop exceeds the safety standard. In this way, according to the classification results, it is helpful to realize real-time monitoring and early warning of smoke in the welding workshop, thereby protecting the health of welders, reducing the risk of occupational diseases, and improving the safety and sustainability of the production environment.

[0006] According to a first aspect of the present application, there is provided a welding fume intelligent monitoring system based on the Internet of Things, which comprises:

[0007] A welding workshop data collection module, which is used to collect smoke concentration values ​​at multiple predetermined time points at various monitoring locations in the welding workshop, environmental data at multiple predetermined time points at various monitoring locations in the welding workshop, and safety standard information on exposure to hazardous substances in the welding workshop;

[0008] A welding workshop feature extraction module, used to extract and analyze the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop, so as to obtain a feature vector of the overall distribution of welding smoke content in the welding workshop and a feature vector of safety standard information;

[0009] A welding workshop feature fusion module, used for fusing the welding fume content overall distribution feature vector and the safety standard information feature vector using a joint encoder to obtain a welding fume safety feature matrix;

[0010] The welding workshop classification result generating module is used to obtain a classification result based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

[0011] In combination with the first aspect of the present application, in an IoT-based welding fume intelligent monitoring system of the first aspect of the present application, the welding workshop feature extraction module includes: a welding fume content analysis unit, which is used to feature encode and comprehensively analyze the fume concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop and the environmental data distribution at multiple predetermined time points at each monitoring position in the welding workshop to obtain a feature vector of the overall distribution of welding fume content in the welding workshop; and a safety standard information analysis unit, which is used to feature encode the safety standard information on exposure to harmful substances in the welding workshop to obtain the safety standard information feature vector.

[0012] According to a second aspect of the present application, a welding fume intelligent monitoring method based on the Internet of Things is provided, which comprises:

[0013] Collect smoke concentration values ​​at multiple predetermined time points at each monitoring location in the welding workshop, environmental data at multiple predetermined time points at each monitoring location in the welding workshop, and safety standard information on exposure to hazardous substances in the welding workshop;

[0014] Extracting and analyzing the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop to obtain a feature vector of the overall distribution of welding smoke content in the welding workshop and a feature vector of safety standard information;

[0015] Using a joint encoder, the welding fume content overall distribution feature vector and the safety standard information feature vector are merged to obtain a welding fume safety feature matrix;

[0016] A classification result is obtained based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

[0017] In combination with the second aspect of the present application, in an Internet of Things-based intelligent welding fume monitoring method of the second aspect of the present application, a classification result is obtained based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard, including: a welding fume safety feature extraction unit, used to pass the welding fume safety feature matrix through a welding fume safety feature extraction module based on a convolutional neural network model to obtain a welding fume safety classification feature vector; an optimization factor acquisition unit, used to perform backward correlation inference on the welding fume safety feature vector based on category probability significance enhancement to obtain an optimization factor; a welding fume safety feature optimization unit, used to weight the welding fume safety feature vector with the optimization factor as a weight to obtain an optimized welding fume safety feature vector; a welding fume safety feature classification unit, used to pass the optimized welding fume safety classification feature vector through a classifier to obtain the classification result.

[0018] Compared with the prior art, the present application provides an intelligent monitoring system and method for welding fume based on the Internet of Things. First, dust concentration detectors, temperature and humidity sensors, and wind speed sensors are installed at various monitoring positions in the welding workshop to collect smoke concentration values ​​at multiple predetermined time points at various monitoring positions in the welding workshop, environmental data at multiple predetermined time points at various monitoring positions in the welding workshop, and safety standard information on exposure to harmful substances in the welding workshop; then, machine learning technology is used to extract and analyze these input data to obtain classification results, and the classification results are used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard. In this way, according to the classification results, it is helpful to realize real-time monitoring and early warning of welding workshop smoke, thereby protecting the health of welders, reducing the risk of occupational diseases, and improving the safety and sustainability of the production environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 The figure illustrates a schematic block diagram of an IoT-based welding fume intelligent monitoring system according to an embodiment of the present application.

[0021] Figure 2 The figure illustrates a schematic block diagram of a welding workshop feature extraction module in an IoT-based welding fume intelligent monitoring system according to an embodiment of the present application.

[0022] Figure 3 The figure illustrates a schematic block diagram of a welding fume content analysis unit in a welding workshop feature extraction module in an IoT-based welding fume intelligent monitoring system according to an embodiment of the present application.

[0023] Figure 4 The figure illustrates a schematic block diagram of a safety standard information analysis unit in a welding workshop feature extraction module in an IoT-based welding fume intelligent monitoring system according to an embodiment of the present application.

[0024] Figure 5 The figure illustrates a schematic block diagram of a welding workshop classification result generation module in an IoT-based welding fume intelligent monitoring system according to an embodiment of the present application.

[0025] Figure 6 The figure illustrates a flow chart of an intelligent welding fume monitoring method based on the Internet of Things according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0027] Figure 1 The figure shows a schematic block diagram of a welding fume intelligent monitoring system based on the Internet of Things according to an embodiment of the present application. Figure 1As shown, according to the embodiment of the present application, the intelligent welding fume monitoring system 100 based on the Internet of Things includes: a welding workshop data acquisition module 110, which is used to collect the fume concentration values ​​of multiple predetermined time points at each monitoring position in the welding workshop, the environmental data of multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on the exposure of harmful substances in the welding workshop; a welding workshop feature extraction module 120, which is used to extract and analyze the fume concentration values ​​of multiple predetermined time points at each monitoring position in the welding workshop, the environmental data of multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on the exposure of harmful substances in the welding workshop to obtain the overall distribution feature vector of the welding fume content in the welding workshop and the safety standard information feature vector; a welding workshop feature fusion module 130, which is used to use a joint encoder to fuse the overall distribution feature vector of the welding fume content and the safety standard information feature vector to obtain a welding fume safety feature matrix; a welding workshop classification result generation module 140, which is used to obtain a classification result based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

[0028] In welding workshops, a large amount of dust and smoke is generated, mainly from welding processes such as spot welding, CO2 gas shielded welding and argon arc welding. In particular, CO2 gas shielded welding and argon arc welding will release a large amount of welding smoke and arc light. In other words, in welding production, a large amount of harmful metal smoke, electric welding dust and toxic gases will be released. When the concentration of these substances exceeds the safety standard, it may cause harm to the health of welders and lead to occupational diseases such as pneumoconiosis, manganese poisoning and metal fever.

[0029] At present, the methods for monitoring welding smoke include chemical analysis, optical methods, etc. Chemical analysis usually requires sample collection for laboratory testing or analysis, and cannot be monitored in real time; while optical methods are sensitive to environmental conditions, such as light intensity, temperature, humidity and other factors that may affect the accuracy of the monitoring results.

[0030] Therefore, we look forward to an intelligent welding fume monitoring system and method based on the Internet of Things, which uses machine learning technology to analyze the fume situation in the welding workshop, so as to achieve real-time monitoring and early warning of the fume in the welding workshop, thereby protecting the health of welders, reducing the risk of occupational diseases, and improving the safety and sustainability of the production environment.

[0031] In the embodiment of the present application, the welding workshop data acquisition module 110 is used to collect smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and safety standard information on exposure to harmful substances in the welding workshop. It should be understood that during the welding process, the smoke concentration will fluctuate with time and process changes. Collecting data at multiple time points can more comprehensively reflect the dynamic changes in smoke concentration. In addition, there may be large differences in smoke concentration at different positions, and a single position cannot represent the situation of the entire workshop. Collecting data at multiple positions can more comprehensively reflect the distribution of smoke inside the workshop. Therefore, first collect the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop as input data. Next, considering that environmental factors such as temperature, humidity and wind speed will affect the diffusion and sedimentation of smoke. Collecting environmental data at the same time can help the system better understand the influence of these environmental factors on smoke concentration, thereby ensuring the accuracy of smoke concentration collection. Therefore, it is also necessary to collect environmental data (temperature value, humidity value and wind speed value) at multiple predetermined time points at each monitoring position in the welding workshop. Furthermore, it should be understood that the safety standard information on exposure to hazardous substances in welding workshops issued by organizations such as the International Labor Organization (ILO) and national occupational safety and health management agencies can be used as a standard for judging whether the smoke concentration in the welding workshop exceeds the standard. Comparing and analyzing the safety standard information with the real-time monitored smoke concentration data can timely detect the exceeding of the standard and provide a basis for safety management decisions. Therefore, the safety standard information on exposure to hazardous substances in welding workshops is further collected as input data.

[0032] In the embodiment of the present application, the welding workshop feature extraction module 120 is used to extract and analyze the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop to obtain the overall distribution feature vector of welding smoke content in the welding workshop and the safety standard information feature vector. It should be understood that after collecting these input data, the input data are further subjected to feature extraction and analysis tasks.

[0033] Specifically, Figure 2 The figure shows a schematic block diagram of a welding workshop feature extraction module in the welding fume intelligent monitoring system based on the Internet of Things according to an embodiment of the present application. Figure 2As shown, the welding workshop feature extraction module 120 includes: a welding fume content analysis unit 121, which is used to feature encode and comprehensively analyze the fume concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop and the environmental data distribution at multiple predetermined time points at each monitoring position in the welding workshop to obtain a feature vector of the overall distribution of welding fume content in the welding workshop; a safety standard information analysis unit 122, which is used to feature encode the safety standard information on exposure to harmful substances in the welding workshop to obtain the safety standard information feature vector.

[0034] First, feature extraction and analysis operations are performed on the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop and the environmental data distribution at multiple predetermined time points at each monitoring position in the welding workshop. Specifically, Figure 3 The figure shows a schematic block diagram of a welding fume content analysis unit in a welding workshop feature extraction module in a welding fume intelligent monitoring system based on the Internet of Things according to an embodiment of the present application. Figure 3 As shown, the welding fume content analysis unit 121 includes: a dust concentration time series feature extraction subunit 121-1, which is used to pass the dust concentration values ​​of multiple predetermined time points at each monitoring position in the welding workshop through a dust concentration time series feature extraction module based on a time series encoder to obtain a dust concentration time series feature vector of each monitoring position; an environmental factor feature extraction subunit 121-2, which is used to arrange the environmental data of multiple predetermined time points at each monitoring position in the welding workshop into multiple environmental factor input vectors for each monitoring position in the welding workshop according to the time dimension, and then pass the environmental factor feature extraction module based on a one-dimensional convolution kernel to obtain multiple environmental factor feature vectors for each monitoring position; an environmental factor time series feature extraction subunit 121-3, which is used to pass the environmental factor time series feature extraction module based on a one-dimensional convolution kernel to obtain multiple environmental factor feature vectors for each monitoring position. The environmental factor feature vectors are passed through an environmental time series change feature extraction module based on a bidirectional long short-term memory neural network model to obtain the environmental factor time series change feature vectors of each monitoring position; the welding fume content analysis subunit 121-4 at each position is used to perform feature fusion on the dust concentration time series feature vectors of each monitoring position and the environmental factor time series change feature vectors of each monitoring position to obtain the welding fume content feature vectors of each monitoring position; the welding fume content analysis subunit 121-5 in the welding workshop is used to arrange the welding fume content feature vectors of each monitoring position into an overall distribution input vector of welding fume in the welding workshop, and then pass it through a fume content overall distribution feature extraction module based on a feature encoder to obtain the overall distribution feature vector of welding fume content in the welding workshop.

[0035] It should be understood that the change of smoke concentration over time is often not random, but is related to various factors such as the time, frequency, type of welding operation and ventilation conditions in the workshop. The timing encoder can capture these change patterns. For example, it can identify the increase or decrease trend of smoke concentration in a specific time period (such as at the beginning or end of work). Therefore, in order to better capture the dynamic changes of smoke concentration in the time dimension at each monitoring position in the welding workshop, the timing encoder is used to perform feature extraction and analysis operations on the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop. Specifically, in an embodiment of the present application, the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop are respectively passed through a dust concentration time series feature extraction module based on a timing encoder to obtain a dust concentration time series feature vector for each monitoring position.

[0036] In a specific embodiment of the present application, the dust concentration time series feature extraction subunit 121-1 is used to: arrange the smoke concentration values ​​of multiple predetermined time points at each monitoring location in the welding workshop into a smoke concentration input vector according to the time dimension; perform maximum value-based normalization on the smoke concentration input vector to obtain a normalized smoke concentration input vector; use the fully connected layer of the time series encoder to fully connect the normalized smoke concentration input vector to extract high-dimensional implicit features of the eigenvalues ​​of each position in the normalized smoke concentration input vector; and use the one-dimensional convolutional layer of the time series encoder to one-dimensionally encode the normalized smoke concentration input vector to extract high-dimensional implicit correlation features of the correlation between the eigenvalues ​​of each position in the normalized smoke concentration input vector.

[0037] Next, it should be understood that environmental data includes temperature, humidity, wind speed, etc., and there is a certain correlation between these data. Therefore, in order to better capture the correlation characteristics between environmental data, the environmental data of multiple predetermined time points at each monitoring position in the welding workshop are first arranged according to the time dimension as multiple environmental factor input vectors for each monitoring position in the welding workshop. Among them, each environmental factor input vector in the multiple environmental factor input vectors is composed of a temperature value, a humidity value, and a wind speed value at a predetermined time point. Then, considering that the one-dimensional convolution kernel can effectively capture the pattern within the local time window in the input data, this is very useful for understanding the short-term effects of environmental factors. For example, a sharp change in temperature in a short period of time may have an immediate impact on the smoke concentration. Based on this, the multiple environmental factor input vectors of each monitoring position in the welding workshop are respectively deep convolutionally encoded by an environmental factor feature extraction module based on a one-dimensional convolution kernel to obtain multiple environmental factor feature vectors for each monitoring position.

[0038] In a specific embodiment of the present application, the environmental factor feature extraction subunit 121-2 is used to: use each layer of the environmental factor feature extraction module based on the one-dimensional convolution kernel to perform the following on the input data in the forward pass of the layer: perform convolution processing based on the one-dimensional convolution kernel on the input data to obtain a convolution feature map; perform pooling processing based on the feature matrix on the convolution feature map to obtain a pooled feature map; perform nonlinear activation on the pooled feature map to obtain an activation feature map; wherein the input of the first layer of the environmental factor feature extraction module based on the one-dimensional convolution kernel is a plurality of environmental factor input vectors of each monitoring position in the welding workshop, and the output of the last layer of the environmental factor feature extraction module based on the one-dimensional convolution kernel is a plurality of environmental factor feature vectors of each monitoring position.

[0039] Next, it should be understood that environmental data (temperature, humidity, wind speed) changes over time, and these changes are crucial to understanding the environmental status of the welding workshop and its impact on smoke concentration. Therefore, in order to better capture environmental characteristics, it is also necessary to analyze the dynamic change characteristics of welding data in the time dimension. Specifically, in an embodiment of the present application, the multiple environmental factor feature vectors of each monitoring position are passed through an environmental time series change feature extraction module based on a bidirectional long short-term memory neural network model to obtain the environmental factor time series change feature vectors of each monitoring position. Among them, the bidirectional long short-term memory neural network model (LSTM: Long Short-Term Memory) is a special recurrent neural network (RNN), which designs a gating mechanism to solve the problem of gradient vanishing or explosion of traditional RNN when processing long sequence data. This enables LSTM to remember information over a longer time span, which is very effective for processing long-term dependencies in time series data.

[0040] In a specific embodiment of the present application, the environmental factor time series feature extraction subunit 121-3 is used to: arrange the multiple environmental factor feature vectors of each monitoring location in one dimension to obtain a sequence of long-distance correlation feature vectors of each monitoring location; and use the environmental time series change feature extraction module based on the bidirectional long short-term memory neural network model to perform long-distance correlation feature encoding on the sequence of long-distance correlation feature vectors of each monitoring location to obtain the environmental factor time series change feature vector of each monitoring location.

[0041] Furthermore, it should be understood that there are close correlation characteristics between the dust concentration in the welding workshop and environmental factors (such as temperature, humidity and wind speed), which can significantly affect the behavior of dust and smoke, including its generation, diffusion and deposition. Therefore, in order to capture these correlation characteristics, the dust concentration time series feature vectors of each monitoring position and the time series change feature vectors of the environmental factors of each monitoring position are respectively fused to obtain the welding fume content feature vectors of each monitoring position.

[0042] Then, considering that a single analysis of the smoke concentration and environmental factors at each monitoring position in the welding workshop is difficult to reflect the overall situation of the entire welding workshop, based on this, the welding smoke content feature vectors of each monitoring position are arranged as the overall distribution input vector of welding smoke in the welding workshop. In this way, a global view of the smoke distribution of the entire workshop can be constructed. Further, considering that the feature encoder is good at automatically extracting hidden patterns and structures from the original data, this is very helpful for identifying complex patterns of smoke distribution, such as periodic changes, seasonal trends, etc. Therefore, the overall distribution feature extraction module of the smoke content based on the feature encoder is used to perform deep convolution encoding on the overall distribution input vector of welding smoke in the welding workshop to obtain the overall distribution feature vector of welding smoke content in the welding workshop. In this way, it not only provides a comprehensive and detailed analysis perspective for the intelligent monitoring of smoke in the welding workshop, but also greatly enhances the prediction, early warning and decision support capabilities of the system.

[0043] In a specific embodiment of the present application, the welding workshop welding fume content analysis subunit 121-5 is used to: use the layers of the smoke content overall distribution feature extraction module based on the feature encoder to respectively perform the following operations on the input data in the forward transfer of the layer: use the convolution units of the layers of the smoke content overall distribution feature extraction module based on the feature encoder to perform convolution processing based on a one-dimensional convolution kernel on the input data to obtain a convolution feature map; use the pooling units of the layers of the smoke content overall distribution feature extraction module based on the feature encoder to perform global mean pooling processing on the convolution feature map to obtain a pooling feature map; and use the activation units of the layers of the smoke content overall distribution feature extraction module based on the feature encoder to perform nonlinear activation on the feature values ​​of each position in the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the smoke content overall distribution feature extraction module based on the feature encoder is the overall distribution feature vector of welding fume content in the welding workshop.

[0044] Then, the safety standard information on exposure to hazardous substances in the welding workshop is extracted and analyzed. Specifically, Figure 4The figure shows a schematic block diagram of the safety standard information analysis unit in the welding workshop feature extraction module in the welding fume intelligent monitoring system based on the Internet of Things according to an embodiment of the present application. Figure 4 As shown, the safety standard information analysis unit 122 includes: a safety standard information embedding coding subunit 122-1, used to embed and encode the safety standard information on exposure to hazardous substances in a welding workshop to obtain a safety standard information embedding vector; a safety standard information semantic understanding subunit 122-2, used to pass the safety standard information embedding vector through a safety standard information feature extraction module based on a context encoder to obtain the safety standard information feature vector.

[0045] It should be understood that safety standard information is usually categorical or numerical data, and directly extracting features from it may cause difficulties in model processing due to inconsistent data types. Embedded coding can convert these standard information into vector form, so that it is in the same data space as other features, which is convenient for unified model processing. Therefore, firstly, the safety standard information on exposure to hazardous substances in welding workshops is embedded and encoded to obtain the safety standard information embedded vector.

[0046] In a specific embodiment of the present application, the safety standard information embedding encoding subunit 122-1 is used to: perform word segmentation processing on the safety standard information on exposure to hazardous substances in a welding workshop to obtain a word sequence; and map the word sequence into an embedding vector to obtain the safety standard information embedding vector.

[0047] Next, considering that safety standard information often contains multiple parts, such as the names of different hazardous substances, concentration limits, applicable scopes, etc., there are inherent logics and connections between these information. The context encoder can capture the relationships between these information, and understand that the meaning of a standard limit is not limited to the value itself, but also includes the context in which it is located, such as the type of specific hazardous substances, monitoring conditions, etc. Therefore, in order to better perform feature analysis on safety standard information, the context encoder is used to understand and analyze the contextual semantics of safety standard information. Specifically, in an embodiment of the present application, the safety standard information embedding vector is passed through a safety standard information feature extraction module based on a context encoder to obtain the safety standard information feature vector.

[0048] In a specific embodiment of the present application, the safety standard information semantic understanding subunit 122-2 is used to: use the converter-based Bert model of the context encoder to perform global context semantic encoding on the security standard information embedding vector to obtain the security standard information feature vector.

[0049] In an embodiment of the present application, the welding workshop feature fusion module 130 is used to use a joint encoder to fuse the welding fume content overall distribution feature vector and the safety standard information feature vector to obtain a welding fume safety feature matrix. It should be understood that the overall distribution feature vector of the fume content provides detailed information about the distribution of fume concentration in the welding workshop, while the safety standard information feature vector clarifies the safety limit of each harmful substance. Fusion of the two can achieve a comprehensive assessment of whether the fume concentration in the workshop exceeds the standard, ensuring the accuracy of the monitoring results. Therefore, in order to more accurately monitor the welding fume in the welding workshop, a joint encoder is used to fuse the overall distribution feature vector of the welding fume content and the safety standard information feature vector to obtain a welding fume safety feature matrix. Among them, the joint encoder can integrate data from different sources and different types into a unified feature representation, which means that for the machine learning model, the actual situation of the fume concentration distribution and the constraints of the safety standards can be learned at the same time.

[0050] In a specific embodiment of the present application, the welding workshop feature fusion module 130 is used to: use the joint encoder to jointly encode the welding fume content overall distribution feature vector and the safety standard information feature vector using the following formula to generate the welding fume safety feature matrix; wherein the formula is:

[0051]

[0052] in Represents vector multiplication, M a represents the welding fume safety characteristic matrix, V a1 represents the overall distribution characteristic vector of welding fume content, V a2 represents the safety standard information feature vector, represents the transpose of the safety standard information feature vector.

[0053] In the embodiment of the present application, the welding workshop classification result generation module 140 is used to obtain a classification result based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard. Specifically, Figure 5 The figure shows a schematic block diagram of a welding workshop classification result generation module in the welding fume intelligent monitoring system based on the Internet of Things according to an embodiment of the present application. Figure 5As shown, the welding workshop classification result generation module 140 includes: a welding fume safety feature extraction unit 141, which is used to pass the welding fume safety feature matrix through a welding fume safety feature extraction module based on a convolutional neural network model to obtain a welding fume safety classification feature vector; an optimization factor acquisition unit 142, which is used to perform backward correlation inference on the welding fume safety feature vector based on category probability significance enhancement to obtain an optimization factor; a welding fume safety feature optimization unit 143, which is used to weight the welding fume safety feature vector with the optimization factor as a weight to obtain an optimized welding fume safety feature vector; a welding fume safety feature classification unit 144, which is used to pass the optimized welding fume safety classification feature vector through a classifier to obtain the classification result.

[0054] It should be understood that although feature fusion has integrated the information of multiple data sources to form a preliminary feature representation, these features may still contain a lot of redundant information or be not refined enough. Through further feature extraction, more abstract, compact and expressive features can be learned, which can often better capture the essential attributes and patterns of the data. Based on this, a welding fume safety feature extraction module based on a convolutional neural network model is used to perform a deeper feature extraction operation on the welding fume safety feature matrix to obtain a welding fume safety classification feature vector. Among them, the convolutional neural network model (CNN) can effectively capture the local features and spatial structure in the data through local perception and weight sharing mechanisms. For the welding fume safety feature matrix, CNN can identify the local patterns of fume concentration distribution and the relationship between these patterns and safety standards, which is crucial to understanding the complexity and subtle changes of fume distribution.

[0055] In particular, in the technical solution of the present application, the smoke concentration value and environmental data may come from different monitoring equipment and locations, with different dimensions and magnitudes, which may lead to inconsistent distribution of feature vectors in feature space. The extraction of dust concentration time series features and environmental factor time series change features requires accurate capture of the change pattern of time series data. If the time series encoder and the bidirectional long short-term memory neural network model fail to effectively capture these subtle changes, the distribution of time series feature vectors may be inconsistent. The one-dimensional convolution kernel and the feature encoder use different algorithms or parameters to extract environmental factor features and the overall distribution features of smoke content. This difference may lead to inconsistent distribution of feature matrices, which in turn leads to poor manifold geometric consistency of the overall feature distribution of the welding smoke safety feature vector. If the distribution of the welding smoke safety feature vector in high-dimensional space is inconsistent, the classifier may find it difficult to learn the mapping relationship between stable features and class labels during training, resulting in a decrease in the performance of the classifier. The inconsistency of the welding smoke safety feature vector may lead to a high uncertainty in the classification result in some cases, affecting the accurate judgment of whether the welding smoke concentration exceeds the standard. Based on this, the welding smoke safety feature vector is subjected to backward correlation inference based on the enhanced significance of the class probability.

[0056] In a specific embodiment of the present application, the optimization factor acquisition unit 142 includes: multiplying the welding fume safety feature vector with the classification weight matrix of the classifier to obtain a welding fume safety weight feature vector; cascading the welding fume safety feature vector and the welding fume safety weight feature vector to obtain a welding fume safety feature-weight information joint vector; passing the welding fume safety feature-weight information joint vector through a first fully connected layer and then through a sigmoid function to obtain a first activation output value; adding the welding fume safety feature vector and the welding fume safety weight feature vector by position to obtain a welding fume safety feature-weight information sum vector; passing the welding fume safety feature-weight information sum vector through a second fully connected layer and then through a sigmoid function to obtain a second activation output value; calculate the mean of the first activation output value and the second activation output value, and subtract 1 from the mean to obtain a first weighting coefficient; calculate the mean of the first activation output value and the second activation output value to obtain a second weighting coefficient; calculate the natural exponential function value with the eigenvalue of the welding fume safety weight feature vector as a power to obtain a first exponential welding fume safety weight feature vector; calculate the natural exponential function value with the eigenvalue of the welding fume safety feature vector as a power to obtain a second exponential welding fume safety feature vector; based on the first weighting coefficient and the second weighting coefficient, calculate the weighted sum of the first exponential welding fume safety weight feature vector and the second exponential welding fume safety feature vector to obtain a weighted sum feature vector, and calculate the second norm of the weighted sum feature vector to obtain an optimization factor.

[0057] The optimization factor acquisition unit 142 further includes: performing backward correlation inference based on category probability significance enhancement on the welding fume safety feature vector using the following formula, wherein the formula is:

[0058]

[0059] Among them, v c represents the welding fume safety feature vector, M represents the classification weight matrix of the classifier, represents matrix multiplication, represents positional addition, concat represents the cascade function, W1 represents the first weight matrix of the first fully connected layer, b1 represents the first bias vector of the first fully connected layer, t1 represents the first activation output value, W2 represents the second weight matrix of the second fully connected layer, b2 represents the second bias vector of the second fully connected layer, t2 represents the second activation output value, sigmoid represents the S-shaped function, ||·||2 represents the second norm of the vector, and α represents the optimization factor.

[0060] That is, in the technical solution of the present application, the manifold geometric consistency of the overall feature distribution of the welding fume safety feature vector is poor, resulting in poor regression norm of the class label of the classifier when it is classified and regressed through the classifier, affecting the accuracy of the classification result. Therefore, in the technical solution of the present application, the welding fume safety feature vector is subjected to backward correlation inference based on enhanced category probability significance, which is performed by using the classification weight matrix of the classifier to perform auxiliary analysis of the feature attribute layout of the welding fume safety feature vector, and the statistical norm of the differential features between the auxiliary analysis result and the original welding fume safety feature vector in the class high-dimensional space is used to perform backward correlation prediction, so as to improve the description significance of the improved welding fume safety feature vector for the category probability of the predetermined category of the classifier. In this way, maintaining the geometric consistency of the welding fume safety feature vector in the process of migration transformation to the class label regression domain of the classifier not only helps to reduce the regression bias in the classification process, but also can improve the generalization ability of the classifier to new data, and finally achieve more accurate classification results.

[0061] Next, it should be understood that the classifier can distinguish between normal and abnormal situations through complex mathematical models, reduce false alarms caused by environmental factors (such as temperature and humidity changes) or missed alarms caused by improper monitoring threshold settings, and improve the accuracy and reliability of the system. Therefore, the classifier is used to pass the optimized welding fume safety classification feature vector through the classifier to obtain a classification result, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

[0062] In summary, the IoT-based intelligent welding fume monitoring system according to the embodiment of the present application is explained, which first installs dust concentration detectors, temperature and humidity sensors, and wind speed sensors at various monitoring positions in the welding workshop, collects smoke concentration values ​​at multiple predetermined time points at various monitoring positions in the welding workshop, environmental data at multiple predetermined time points at various monitoring positions in the welding workshop, and safety standard information on exposure to harmful substances in the welding workshop; then uses machine learning technology to extract and analyze these input data to obtain classification results, and the classification results are used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard. In this way, according to the classification results, it is helpful to realize real-time monitoring and early warning of welding workshop smoke, thereby protecting the health of welders, reducing the risk of occupational diseases, and improving the safety and sustainability of the production environment.

[0063] As described above, the welding fume intelligent monitoring system 100 based on the Internet of Things according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for welding fume intelligent monitoring based on the Internet of Things. In one example, the welding fume intelligent monitoring system 100 based on the Internet of Things according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the welding fume intelligent monitoring system 100 based on the Internet of Things can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the welding fume intelligent monitoring system 100 based on the Internet of Things can also be one of the many hardware modules of the wireless terminal.

[0064] Alternatively, in another example, the IoT-based welding fume intelligent monitoring system 100 and the wireless terminal may also be separate devices, and the IoT-based welding fume intelligent monitoring system 100 may be connected to the wireless terminal via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0065] Figure 6 The figure shows a flow chart of a welding fume intelligent monitoring method based on the Internet of Things according to an embodiment of the present application. Figure 6As shown, the IoT-based intelligent welding fume monitoring method according to the embodiment of the present application includes: S1, collecting smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and safety standard information on exposure to harmful substances in the welding workshop; S2, respectively extracting and analyzing the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop to obtain a general distribution feature vector of the welding fume content in the welding workshop and a safety standard information feature vector; S3, using a joint encoder to fuse the general distribution feature vector of the welding fume content and the safety standard information feature vector to obtain a welding fume safety feature matrix; S4, obtaining a classification result based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

[0066] Here, those skilled in the art can understand that the specific functions and operations of each step in the above-mentioned welding fume intelligent monitoring method based on the Internet of Things have been referred to above. Figure 1 The description of the IoT-based welding fume intelligent monitoring system has been introduced in detail, and therefore, its repeated description will be omitted.

[0067] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, apparatuses and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0068] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0069] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0070] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0071] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or systems stated in a system claim can also be implemented by one unit or system through software or hardware. Secondary words are used to indicate names, but not to indicate any particular order.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution 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 solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.

Claims

1. An intelligent welding fume monitoring system based on the Internet of Things, characterized in that: include: A welding workshop data collection module, which is used to collect smoke concentration values ​​at multiple predetermined time points at various monitoring locations in the welding workshop, environmental data at multiple predetermined time points at various monitoring locations in the welding workshop, and safety standard information on exposure to hazardous substances in the welding workshop; A welding workshop feature extraction module, used to extract and analyze the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop, so as to obtain a feature vector of the overall distribution of welding smoke content in the welding workshop and a feature vector of safety standard information; A welding workshop feature fusion module, used for fusing the welding fume content overall distribution feature vector and the safety standard information feature vector using a joint encoder to obtain a welding fume safety feature matrix; The welding workshop classification result generating module is used to obtain a classification result based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

2. The welding fume intelligent monitoring system based on the Internet of Things according to claim 1 is characterized in that: The welding workshop feature extraction module comprises: a welding fume content analysis unit, used for performing feature coding and comprehensive analysis on the fume concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop and the distribution of environmental data at multiple predetermined time points at each monitoring position in the welding workshop to obtain a feature vector of the overall distribution of welding fume content in the welding workshop; The safety standard information analysis unit is used to perform feature encoding on the safety standard information on exposure to hazardous substances in a welding workshop to obtain a feature vector of the safety standard information.

3. The welding fume intelligent monitoring system based on the Internet of Things according to claim 2 is characterized in that: The welding fume content analysis unit comprises: A dust concentration time series feature extraction subunit, used for respectively passing the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop through a dust concentration time series feature extraction module based on a time series encoder to obtain a dust concentration time series feature vector at each monitoring position; An environmental factor feature extraction subunit is used to arrange the environmental data of multiple predetermined time points at each monitoring position in the welding workshop into multiple environmental factor input vectors of each monitoring position in the welding workshop according to the time dimension, and then obtain multiple environmental factor feature vectors of each monitoring position through an environmental factor feature extraction module based on a one-dimensional convolution kernel; An environmental factor time series feature extraction subunit is used to obtain an environmental factor time series change feature vector for each monitoring location by passing the multiple environmental factor feature vectors of each monitoring location through an environmental time series change feature extraction module based on a bidirectional long short-term memory neural network model; The welding fume content analysis subunit at each monitoring position is used to perform feature fusion on the dust concentration time series feature vector at each monitoring position and the environmental factor time series change feature vector at each monitoring position to obtain the welding fume content feature vector at each monitoring position; The welding fume content analysis subunit of the welding workshop is used to arrange the welding fume content feature vectors of the various monitoring positions into an overall distribution input vector of welding fume in the welding workshop and then obtain the overall distribution feature vector of welding fume content in the welding workshop through a fume content overall distribution feature extraction module based on a feature encoder.

4. The welding fume intelligent monitoring system based on the Internet of Things according to claim 3 is characterized in that: The dust concentration time series feature extraction subunit is used to: Arranging the smoke concentration values ​​at a plurality of predetermined time points at each monitoring position in the welding workshop according to the time dimension as a smoke concentration input vector; The smoke concentration input vector is normalized based on the maximum value to obtain a normalized smoke concentration input vector; Using the fully connected layer of the temporal encoder to perform fully connected encoding on the normalized smoke concentration input vector to extract high-dimensional implicit features of the feature values ​​at each position in the normalized smoke concentration input vector; as well as The normalized smoke concentration input vector is one-dimensionally encoded using the one-dimensional convolutional layer of the temporal encoder to extract high-dimensional implicit correlation features of the correlation between the feature values ​​at each position in the normalized smoke concentration input vector.

5. The welding fume intelligent monitoring system based on the Internet of Things according to claim 4 is characterized in that: The safety standard information analysis unit comprises: A safety standard information embedding and encoding subunit, used for embedding and encoding the safety standard information on exposure to hazardous substances in a welding workshop to obtain a safety standard information embedding vector; The safety standard information semantic understanding subunit is used to embed the safety standard information into a vector and pass it through a safety standard information feature extraction module based on a context encoder to obtain the safety standard information feature vector.

6. The welding fume intelligent monitoring system based on the Internet of Things according to claim 5 is characterized in that: The welding workshop feature fusion module is used to: use the joint encoder to jointly encode the welding fume content overall distribution feature vector and the safety standard information feature vector according to the following formula to generate the welding fume safety feature matrix; Wherein, the formula is: in Represents vector multiplication, M a represents the welding fume safety characteristic matrix, V a1 represents the overall distribution characteristic vector of welding fume content, V a2 represents the safety standard information feature vector, represents the transpose of the safety standard information feature vector.

7. The welding fume intelligent monitoring system based on the Internet of Things according to claim 6 is characterized in that: The welding workshop classification result generation module includes: A welding fume safety feature extraction unit, used for passing the welding fume safety feature matrix through a welding fume safety feature extraction module based on a convolutional neural network model to obtain a welding fume safety classification feature vector; An optimization factor acquisition unit, used for performing backward correlation inference on the welding fume safety feature vector based on category probability significance enhancement to obtain an optimization factor; a welding fume safety feature optimization unit, configured to weight the welding fume safety feature vector using the optimization factor as a weight to obtain an optimized welding fume safety feature vector; The welding fume safety feature classification unit is used to pass the optimized welding fume safety classification feature vector through a classifier to obtain the classification result.

8. The welding fume intelligent monitoring system based on the Internet of Things according to claim 7 is characterized in that: The optimization factor acquisition unit includes: Multiplying the welding fume safety feature vector by a classification weight matrix of a classifier to obtain a welding fume safety weight feature vector; Cascading the welding fume safety feature vector and the welding fume safety weight feature vector to obtain a welding fume safety feature-weight information joint vector; The welding fume safety feature-weight information joint vector passes through the first fully connected layer and then through the sigmoid function to obtain a first activation output value; Add the welding fume safety feature vector and the welding fume safety weight feature vector according to position to obtain a welding fume safety feature-weight information sum vector; The welding fume safety feature-weight information sum vector passes through the second fully connected layer and then passes through the sigmoid function to obtain a second activation output value; Calculating a mean of the first activation output value and the second activation output value, and subtracting the mean from 1 to obtain a first weighting coefficient; Calculating a mean of the first activation output value and the second activation output value to obtain a second weighting coefficient; Calculating a natural exponential function value with the eigenvalue of the welding fume safety weight eigenvector as a power to obtain a first exponential welding fume safety weight eigenvector; Calculating a natural exponential function value with the eigenvalue of the welding fume safety feature vector raised to a power to obtain a second exponential welding fume safety feature vector; Based on the first weighting coefficient and the second weighting coefficient, a weighted sum of the first exponential welding fume safety weight feature vector and the second exponential welding fume safety feature vector is calculated to obtain a weighted sum feature vector, and a bi-norm of the weighted sum feature vector is calculated to obtain an optimization factor.

9. An intelligent monitoring method for welding smoke based on the Internet of Things, characterized in that ,include: Collect smoke concentration values ​​at multiple predetermined time points at each monitoring location in the welding workshop, environmental data at multiple predetermined time points at each monitoring location in the welding workshop, and safety standard information on exposure to hazardous substances in the welding workshop; Extracting and analyzing the smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop to obtain a feature vector of the overall distribution of welding smoke content in the welding workshop and a feature vector of safety standard information; Using a joint encoder, the welding fume content overall distribution feature vector and the safety standard information feature vector are merged to obtain a welding fume safety feature matrix; A classification result is obtained based on the welding fume safety feature matrix, and the classification result is used to indicate whether the welding fume concentration in the welding workshop exceeds the safety standard.

10. The method for intelligent monitoring of welding smoke based on the Internet of Things according to claim 9 is characterized in that: The smoke concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop, the environmental data at multiple predetermined time points at each monitoring position in the welding workshop, and the safety standard information on exposure to harmful substances in the welding workshop are respectively extracted and analyzed to obtain the overall distribution feature vector of the welding smoke content in the welding workshop and the safety standard information feature vector, including: a welding fume content analysis unit, used for performing feature coding and comprehensive analysis on the fume concentration values ​​at multiple predetermined time points at each monitoring position in the welding workshop and the distribution of environmental data at multiple predetermined time points at each monitoring position in the welding workshop to obtain a feature vector of the overall distribution of welding fume content in the welding workshop; The safety standard information analysis unit is used to perform feature encoding on the safety standard information on exposure to hazardous substances in a welding workshop to obtain a feature vector of the safety standard information.

Citation Information

Cited By

  • Dynamic prediction and optimization calculation method for high manganese steel welding fume emission

    CN120297520A

  • High-altitude welding robot system based on environment self-adaptive control

    CN121670685A

  • A high-altitude welding robot system based on environment adaptive control

    CN121670685B

  • Working condition perceived wire harness welding smoke capture and tin powder recovery self-evolution control method

    CN122592901A