Safety monitoring method and system for paint production workshop

Through the combination of multi-level, multi-dimensional data acquisition and joint training models, all-round safety monitoring of the paint workshop is achieved, the problem of incomplete safety monitoring in the existing technology is solved, and the accuracy of abnormal identification and safety accident prevention capabilities are improved.

CN119087888AActive Publication Date: 2024-12-06JIANGXI YAKA TECH CO LTD
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411371689.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-29
Publication Date
2024-12-06
Estimated Expiration
2044-09-29

AI Technical Summary

Technical Problem

The existing technology lacks a comprehensive management system in the safety monitoring of paint workshops, fails to effectively prevent safety incidents, and is inaccurate in the analysis of safety data, and fails to identify and optimize in combination with the connection between data.

Method used

Using multi-level and multi-dimensional data acquisition and real-time detection, through pre-trained identification models and joint training real-time monitoring models and prevention models, all-round safety monitoring of workers, equipment, environment and production quality is carried out, combined with regional and correlation control Early warning policy library to realize automatic control and early warning.

Benefits of technology

It improves the accuracy of abnormal identification, enhances the ability to prevent safety accidents, improves the processing efficiency and accuracy through automatic control and early warning, and predicts potential risks through time series data, improving the intelligence level of safety management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119087888A_ABST
    Figure CN119087888A_ABST
Patent Text Reader

Abstract

The invention discloses a safety monitoring method and system for a paint production workshop, and relates to the technical field of safety monitoring. A safety monitoring system for a paint production workshop comprises a data acquisition module, a real-time monitoring module, a prevention module, a control module, an early warning module and a recording module. According to the invention, comprehensive safety monitoring of workers, equipment, environment and production quality in a workshop is realized through multi-dimensional acquisition of sensor data, sound data and image data in combination with a pre-trained identification model; through the pre-trained identification model and the real-time monitoring model and the prevention model which are jointly trained, the accuracy of anomaly identification is improved, abnormal conditions under different security categories can be identified, and the accuracy of anomaly detection is enhanced through optimization training; through the regional control early warning strategy library and the association control early warning strategy library, automatic control and early warning of the workshop are realized according to the monitoring result and the prediction result, and the processing efficiency and accuracy are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of safety monitoring, and in particular to a safety monitoring method and system for a paint production workshop. Background Art

[0002] Paint plays an important role in industrial production and is widely used in automobile manufacturing, furniture manufacturing, home appliance manufacturing and other industries. However, there are many potential risk factors in paint production workshops, including the release of volatile organic compounds (VOCs), the accumulation of flammable and explosive gases, the high temperature operation of production equipment and the safety hazards of workers. Especially in the context of efficient painting operations, how to ensure the safety of workers, equipment, environment and production quality is particularly critical.

[0003] In the existing technology, the monitoring methods for paint workshop safety are mostly concentrated in one aspect, such as simple environmental monitoring or equipment fault detection, and a complete safety management system has not been formed; only real-time judgment is made on the data that has been generated, and there is a lack of prevention capabilities for workshop safety; there are logs for safety incidents, but there is a lack of intelligent analysis of the records, and the records and safety monitoring are not connected and optimized with each other; the quality of the paint production process cannot be intelligently identified; the analysis of safety data is not combined with the connection between the data for identification and optimization, and the analysis results are inaccurate. Summary of the invention

[0004] In view of the deficiencies of the above-mentioned prior art, the present invention proposes a comprehensive paint shop safety monitoring method and system, which ensures all-round safety monitoring of workers, equipment, environment and production quality through multi-level and multi-dimensional data collection, real-time detection, preventive control, early warning recording and other modules, organically integrates worker safety, equipment safety, environmental safety and production quality safety, and ensures safe and efficient production in the paint shop through intelligent monitoring means.

[0005] A safety monitoring method for a paint production workshop comprises the following steps: Step S1: obtaining sensor data, sound data and image data of each area of ​​the paint workshop, sending the sound data and image data to a pre-trained recognition model for abnormality recognition, obtaining sound monitoring results and image monitoring results, using the sensor data, sound monitoring results and image monitoring results as safety data, pre-processing the safety data, and adding safety monitoring tags to the safety data; the safety monitoring tags include safety category, data collection area, collection time and data type, and the safety categories include worker safety, equipment safety, environmental safety and production quality safety; Step S2: Send the safety data acquired in real time into the real-time monitoring model one by one for calculation, verify the real-time monitoring results according to the prediction results, and obtain the real-time monitoring results; optimize the real-time monitoring model and the prevention model through joint training, so that the results of the models can be verified between the models, thereby improving the recognition accuracy of the models; Step S3: Send the acquired time series safety data to the prevention model for calculation, optimize the prediction results according to the real-time monitoring results to obtain the prediction results, match the prediction results to obtain the prediction control results; perform emergency treatment for the emergency situations in both the real-time monitoring results and the prediction control results; save the processing results of the safety data and generate a safety training report; Step S4: divide the real-time monitoring results and the predictive control results according to the data collection area and the safety category; match the regional control strategy and the regional early warning strategy in the regional control and early warning strategy library according to the real-time monitoring results and the predictive control results in the same area, and automatically control and warn the workshop; Step S5: matching the associated control strategy and the associated warning strategy in the associated control warning strategy library according to the real-time monitoring results and the predictive control results of the same safety category, and performing automatic control and warning; Step S6: Record the safety data and processing results in the workshop. The records will be used to optimize the real-time monitoring model, prevention model, regional control warning strategy library and associated control warning strategy library.

[0006] As a preferred technical solution of the present invention, the pre-trained recognition model includes: Independent recognition model training is performed for each source of sound data and image data. Each recognition model is only used for the recognition of corresponding data. The recognition model is built on the basis of support vector machine and is obtained through training and optimization with several annotated historical data. The final output of the pre-trained recognition model is the monitoring result of sound data and image data.

[0007] As a preferred technical solution of the present invention, the real-time monitoring model and the prevention model include: The real-time monitoring model includes: The sensor data processing layer is used to process sensor data, mainly gas concentration sensor data, temperature sensor data and humidity sensor data. The data of each sensor is firstly judged by threshold value to obtain the abnormal weight α, and then the abnormal discrimination value P is obtained according to the change value of the sensor data. The abnormal value K of the sensor is calculated by the abnormal discrimination value and the abnormal weight, and the sensor data processing result is obtained according to the interval where the abnormal value is located. The calculation formula of the abnormal value is:

[0008] Where α1 represents the abnormal weight of the data at the first moment, and P1 represents the sensor change value at the first moment; The real-time monitoring layer is used to classify the sensor data processing results, sound monitoring results and image monitoring results to obtain preliminary real-time monitoring results; The result optimization layer is used to optimize the preliminary real-time monitoring results according to the prediction results to obtain the real-time monitoring results; The prevention model includes: The data merging layer is used to reconstruct the sensor data processing results, sound monitoring results and image monitoring results to obtain time-series safety prediction data; The security prediction layer is used to calculate based on the security prediction data and obtain preliminary prediction results; The result optimization layer is used to optimize the preliminary prediction results according to the real-time monitoring results to obtain the prediction results.

[0009] As a preferred technical solution of the present invention, the joint training of the real-time monitoring model and the prevention model includes: Construct an initial real-time monitoring model and an initial prevention model. The initial real-time monitoring model is constructed based on a deep convolutional neural network, and the initial prevention model is constructed based on a long-short time neural network. The initial real-time monitoring model contains 1 input layer, 1 data calculation layer, 16 hidden layers, and 2 output layers. The initial prevention model contains 1 input layer, 36 hidden layers, and 2 output layers. The last 4 hidden layers of the two models are shared to optimize the results. Obtain some safety data, and annotate real-time abnormal results and prediction results, form the annotated data into a safety data set, and divide the safety data set into a verification set, a test set, and a training set; A certain proportion of data from the safety data training set and validation set is selected to train the initial real-time monitoring model and the initial prevention model, respectively, with real-time abnormal results and prediction results as the targets, to obtain the optimized real-time monitoring model and the optimized prevention model; A certain proportion of data from the security data test set is selected to train the optimized real-time monitoring model and the optimized prevention model. The correct ratio of the output of the final output layer to the actual annotation is taken as the accurate value. The hyperparameters of the two models are adjusted according to the accurate value to obtain the real-time monitoring model and prevention model with the final accuracy rate meeting the standard.

[0010] As a preferred technical solution of the present invention, the regional control warning strategy library and the associated control warning strategy library include: The regional control and early warning strategy library stores the specific control and early warning strategies for the real-time monitoring results and preventive control results in the region; the control and early warning strategies in the regional control and early warning strategy library are more targeted and can control and warn abnormal situations in the current region in the first place; The associated control and early warning strategy library stores the specific control and early warning strategies for all real-time monitoring results and preventive control results of the same security category. Compared with the regional control and early warning strategy library, the associated control and early warning strategy library associates the security data analysis results between different regions and obtains potential abnormal situation handling strategies through the association between regions.

[0011] As a preferred technical solution of the present invention, verifying the real-time monitoring results according to the prediction results includes: After the real-time monitoring model outputs the results, it is verified through the correspondence between the prediction results and the real-time monitoring results in the real-time prediction association table. The real-time prediction association table stores the direct and indirect correspondence between the prediction results and the real-time monitoring results. The direct correspondence indicates that there is a strong correlation between the prediction results and the real-time monitoring results, and the indirect correlation indicates that there is a weak correlation between the prediction results and the real-time monitoring results. If the verification result is a strong correlation, the verification is passed. If the verification result is a weak correlation, multiple consecutive verifications of weak correlation are required to pass the verification. If the frequency of the verification result that there is no correlation exceeds the threshold, it will be recorded, and the record will be used to update the real-time prediction association table.

[0012] As a preferred technical solution of the present invention, generating a safety training report includes: Acquire and process the security data processing results in different time periods, including sensor data, sound data, image data and their corresponding anomaly detection results and emergency processing records; perform cluster analysis on the security data, group and classify similar security events according to the clustering results, and generate clustering results for each type of event; perform keyword extraction on the key features in the clustering results, perform statistical analysis based on the frequency of occurrence of the keywords, and generate report content related to various types of safety hazards; predict possible future security risks based on historical data and the temporal characteristics of security events through time series analysis methods, and incorporate the prediction results into the safety training report; the safety training report includes a graphical display of security data, intuitively presenting the security risks and data distribution in different areas.

[0013] A safety monitoring system for a paint production workshop includes the following modules: The data acquisition module includes a data acquisition unit and a data processing unit. The data acquisition unit is used to obtain sensor data, sound data and image data from various areas of the paint shop. The data processing unit is used to send the sound data and image data to a pre-trained recognition model for abnormality recognition, obtain sound monitoring results and image monitoring results, use the sensor data, sound monitoring results and image monitoring results as safety data, pre-process the safety data, and add safety monitoring tags to the safety data. The real-time monitoring module includes a real-time monitoring unit and a model optimization unit. The real-time monitoring unit contains a real-time monitoring model for calculating the safety data obtained in real time to obtain real-time monitoring results; the model optimization unit is used to optimize the real-time monitoring model and the prevention model, and associate the real-time monitoring model with the prevention model through joint training and record optimization, and use the training set of the record optimization model; The prevention module includes a prediction unit, a prediction control unit, an emergency processing unit and a safety training unit. The prediction unit is used to send the acquired safety data into the prevention model for calculation, and optimize the prediction results according to the real-time monitoring results to obtain the prediction results; the prediction control unit is used to match the prediction control results according to the prediction results; the emergency processing unit is used to process the emergency situations in the real-time monitoring results and the prevention control results; the safety training unit is used to save the processing results of the safety data and generate a safety training report; The early warning module and the control module are integrated in the control early warning module. In addition to the early warning module and the control module, the control early warning module also includes a result division unit and a matching unit. The result division unit is used to divide the real-time monitoring results and the prediction control results according to the data collection area and the safety category; the matching unit is used to match the regional control strategy and the regional early warning strategy in the regional control early warning strategy library according to the real-time monitoring results and the prediction control results in the same area, and match the associated control strategy and the associated early warning strategy in the associated control early warning strategy library according to the real-time monitoring results and the prediction control results in the same safety category; A control module, used for automatically controlling the workshop according to the matched regional control strategy and the associated control strategy; The early warning module is used to automatically warn the workshop according to the matched regional early warning strategy and associated early warning strategy; The recording module is used to record the safety data and processing results in the workshop. The records will be used to optimize the real-time monitoring model, prevention model, regional control and early warning strategy library and associated control and early warning strategy library.

[0014] The present invention has the following advantages: The present invention realizes comprehensive safety monitoring of workers, equipment, environment and production quality in the workshop through multi-dimensional collection of sensor data, sound data and image data, combined with pre-trained recognition models; the accuracy of abnormality recognition is improved through pre-trained recognition models and jointly trained real-time monitoring models and prevention models, and it is able to identify abnormal situations under different safety categories (worker safety, equipment safety, environmental safety, production quality safety), and enhance the accuracy of abnormality detection through optimization algorithms; through the regional control warning strategy library and the associated control warning strategy library, automatic control and warning of the workshop are realized according to the monitoring results and prediction results, thereby improving processing efficiency and accuracy.

[0015] The present invention predicts potential safety risks through time series data and performs emergency processing according to the risk level, thereby improving the ability to prevent safety accidents.

[0016] The present invention continuously improves the performance of real-time monitoring models and prevention models by recording all security data and their processing results and using them for subsequent model training and optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the structure of a safety monitoring system for a paint production workshop used in an embodiment of the present invention; Figure 2 It is a structural diagram of a prevention module of a safety monitoring system for a paint shop in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0019] Embodiment 1, a safety monitoring method for a paint production workshop, comprising the following steps: Step S1: obtaining sensor data, sound data and image data of each area of ​​the paint workshop, sending the sound data and image data to a pre-trained recognition model for abnormality recognition, obtaining sound monitoring results and image monitoring results, using the sensor data, sound monitoring results and image monitoring results as safety data, pre-processing the safety data, and adding safety monitoring tags to the safety data; the safety monitoring tags include safety category, data collection area, collection time and data type, and the safety categories include worker safety, equipment safety, environmental safety and production quality safety; The structural information of the workshop is obtained, and the paint workshop is divided into areas according to the structural information of the workshop. The source of data acquisition, the corresponding data category and the security category to which the data belongs are marked in the area. In the specific implementation process, the structure of the workshop is constructed in 2D or 3D according to the complexity and actual situation of the workshop. The structural diagram of the workshop is used to intuitively display the workshop and the relationship between the various areas of the workshop. The relationship is obtained by the security category to which the data belongs. The pre-trained recognition models include: Independent recognition model training is performed for each source of sound data and image data, and each recognition model is only used for the recognition of corresponding data; the recognition model is built on the basis of support vector machine, and is obtained through training and optimization of several annotated historical data. The final output of the pre-trained recognition model is the monitoring result of sound data and image data; in the specific implementation process, the sound data and image data are used to identify the equipment safety and production quality safety of the equipment in the area, and the workers are identified to ensure their safety. Therefore, in the training data set, for areas where workers may appear, the amount of worker image data in the data set is increased; Step S2: Send the safety data acquired in real time into the real-time monitoring model one by one for calculation, verify the real-time monitoring results according to the prediction results, and obtain the real-time monitoring results; optimize the real-time monitoring model and the prevention model through joint training, so that the results of the models can be verified between the models, thereby improving the recognition accuracy of the models; Step S3: Send the acquired time series safety data to the prevention model for calculation, optimize the prediction results according to the real-time monitoring results to obtain the prediction results, match the prediction results to obtain the prediction control results; perform emergency treatment for the emergency situations in both the real-time monitoring results and the prediction control results; save the processing results of the safety data and generate a safety training report; The real-time monitoring model includes: The sensor data processing layer is used to process sensor data, mainly gas concentration sensor data, temperature sensor data and humidity sensor data. The data of each sensor is firstly judged by threshold value to obtain the abnormal weight α, and then the abnormal discrimination value P is obtained according to the change value of the sensor data. The abnormal value K of the sensor is calculated by the abnormal discrimination value and the abnormal weight, and the sensor data processing result is obtained according to the interval where the abnormal value is located. The calculation formula of the abnormal value is:

[0020] Where α1 represents the abnormal weight of the data at the first moment, and P1 represents the sensor change value at the first moment; The real-time monitoring layer is used to classify the sensor data processing results, sound monitoring results and image monitoring results to obtain preliminary real-time monitoring results; The result optimization layer is used to optimize the preliminary real-time monitoring results according to the prediction results to obtain the real-time monitoring results.

[0021] The prevention model includes: The data merging layer is used to reconstruct the sensor data processing results, sound monitoring results and image monitoring results to obtain time-series safety prediction data; The security prediction layer is used to calculate based on the security prediction data and obtain preliminary prediction results; The result optimization layer is used to optimize the preliminary prediction results according to the real-time monitoring results to obtain the prediction results.

[0022] The joint training of the real-time monitoring model and the prevention model includes: Construct an initial real-time monitoring model and an initial prevention model. The initial real-time monitoring model is constructed based on a deep convolutional neural network, and the initial prevention model is constructed based on a long-short time neural network. The initial real-time monitoring model contains 1 input layer, 1 data calculation layer, 16 hidden layers, and 2 output layers. The initial prevention model contains 1 input layer, 36 hidden layers, and 2 output layers. The last 4 hidden layers of the two models are shared to optimize the results. Obtain some safety data, and annotate real-time abnormal results and prediction results, form the annotated data into a safety data set, and divide the safety data set into a verification set, a test set, and a training set; A certain proportion of data from the safety data training set and validation set is selected to train the initial real-time monitoring model and the initial prevention model, respectively, with real-time abnormal results and prediction results as the targets, to obtain the optimized real-time monitoring model and the optimized prevention model; A certain proportion of data from the security data test set is selected to train the optimized real-time monitoring model and the optimized prevention model. The correct ratio of the output of the final output layer to the actual annotation is used as the accurate value. The hyperparameters of the two models are adjusted according to the accurate value to obtain the real-time monitoring model and prevention model with the final accuracy rate meeting the standard. Verification of real-time monitoring results based on prediction results includes: After the real-time monitoring model outputs the results, it is verified through the correspondence between the prediction results and the real-time monitoring results in the real-time prediction association table. The real-time prediction association table stores the direct and indirect correspondence between the prediction results and the real-time monitoring results. The direct correspondence indicates that there is a strong correlation between the prediction results and the real-time monitoring results, and the indirect correlation indicates that there is a weak correlation between the prediction results and the real-time monitoring results. If the verification result is a strong correlation, the verification is passed. If the verification result is a weak correlation, multiple consecutive verifications of weak correlation are required to pass the verification. If the frequency of the verification result that there is no correlation exceeds the threshold, it will be recorded, and the record will be used to update the real-time prediction association table.

[0023] Generate a safety training report including: Acquire and process the safety data processing results in different time periods, including sensor data, sound data, image data and their corresponding abnormal detection results and emergency processing records; perform cluster analysis on the safety data, group and classify similar safety events according to the clustering results, and generate clustering results for each type of event; perform keyword extraction on the key features in the clustering results, perform statistical analysis based on the frequency of occurrence of the keywords, and generate report content related to various types of safety hazards; predict possible safety risks in the future based on the temporal characteristics of historical data and safety events through time series analysis methods, and incorporate the prediction results into the safety training report; dynamically adjust the report content according to the latest safety policy library and monitoring data, and generate personalized safety training plans for workshops in different areas, including operating specifications and preventive measures for high-frequency abnormal situations in specific areas; the safety training report includes a graphical display of safety data, intuitively presenting the safety risks and data distribution in different areas, and the safety training report can be generated periodically and optimized and adjusted according to the accumulation of historical data and the update of real-time monitoring data to provide safety training suggestions that meet the actual situation of the workshop.

[0024] Step S4: divide the real-time monitoring results and the predictive control results according to the data collection area and the safety category; match the regional control strategy and the regional early warning strategy in the regional control and early warning strategy library according to the real-time monitoring results and the predictive control results in the same area, and automatically control and warn the workshop; Step S5: matching the associated control strategy and the associated warning strategy in the associated control warning strategy library according to the real-time monitoring results and the predictive control results of the same safety category, and performing automatic control and warning; Control and early warning are independent operations. There is a certain correlation between them, but they can be performed independently. The specific correlation is obtained through the strategy library; The regional control warning strategy library and the associated control warning strategy library include: The regional control and early warning strategy library stores the specific control and early warning strategies for the real-time monitoring results and preventive control results in the region; the control and early warning strategies in the regional control and early warning strategy library are more targeted and can control and warn abnormal situations in the current region in the first place; The associated control and early warning strategy library stores the specific control and early warning strategies for all real-time monitoring results and preventive control results of the same security category. Compared with the regional control and early warning strategy library, the associated control and early warning strategy library associates the security data analysis results between different regions and obtains potential abnormal situation handling strategies through the association between regions.

[0025] Step S6: Record the safety data and processing results in the workshop. The records will be used to optimize the real-time monitoring model, prevention model, regional control warning strategy library and associated control warning strategy library.

[0026] The recorded safety data includes original safety data, real-time monitoring results, predictive control results, control strategies and early warning strategies. Incorrect handling of abnormal situations in the saved records will be marked, and records exceeding the rated number will be added to the regional control early warning strategy library and the associated control early warning strategy library. Correct records are used as training data for real-time monitoring models and prevention models to optimize the models.

[0027] Example 2, a safety monitoring system for a paint production workshop, see Figure 1 and Figure 2 As shown, it includes the following modules: The data acquisition module includes a data acquisition unit and a data processing unit. The data acquisition unit is used to obtain sensor data, sound data and image data from various areas of the paint shop. The data processing unit is used to send the sound data and image data to a pre-trained recognition model for abnormality recognition, obtain sound monitoring results and image monitoring results, use the sensor data, sound monitoring results and image monitoring results as safety data, pre-process the safety data, and add safety monitoring tags to the safety data. The real-time monitoring module includes a real-time monitoring unit and a model optimization unit. The real-time monitoring unit contains a real-time monitoring model for calculating the safety data obtained in real time to obtain real-time monitoring results; the model optimization unit is used to optimize the real-time monitoring model and the prevention model, and associate the real-time monitoring model with the prevention model through joint training and record optimization, and use the training set of the record optimization model; The prevention module includes a prediction unit, a prediction control unit, an emergency processing unit and a safety training unit. The prediction unit is used to send the acquired safety data into the prevention model for calculation, and optimize the prediction results according to the real-time monitoring results to obtain the prediction results; the prediction control unit is used to match the prediction control results according to the prediction results; the emergency processing unit is used to process the emergency situations in the real-time monitoring results and the prevention control results; the safety training unit is used to save the processing results of the safety data and generate a safety training report; The early warning module and the control module are integrated in the control early warning module. In addition to the early warning module and the control module, the control early warning module also includes a result division unit and a matching unit. The result division unit is used to divide the real-time monitoring results and the prediction control results according to the data collection area and the safety category; the matching unit is used to match the regional control strategy and the regional early warning strategy in the regional control early warning strategy library according to the real-time monitoring results and the prediction control results in the same area, and match the associated control strategy and the associated early warning strategy in the associated control early warning strategy library according to the real-time monitoring results and the prediction control results in the same safety category; A control module, used for automatically controlling the workshop according to the matched regional control strategy and the associated control strategy; The early warning module is used to automatically warn the workshop according to the matched regional early warning strategy and associated early warning strategy; The recording module is used to record the safety data and processing results in the workshop. The records will be used to optimize the real-time monitoring model, prevention model, regional control and early warning strategy library and associated control and early warning strategy library.

[0028] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A safety monitoring method for a paint production workshop, It is characterized in that The following steps are involved: Step S1: obtaining structural information of the workshop, and dividing the paint workshop into areas according to the structural information of the workshop, obtaining sensor data, sound data and image data of each area of ​​the paint workshop, sending the sound data and image data to a pre-trained recognition model for abnormality recognition, obtaining sound monitoring results and image monitoring results, using the sensor data, sound monitoring results and image monitoring results as security data, pre-processing the security data, and adding security monitoring tags to the security data; Safety monitoring tags include safety category, data collection area, collection time and data type. Safety categories include worker safety, equipment safety, environmental safety and production quality safety. Mark the source of data acquisition, the corresponding data category and the security category to which the data belongs for the workshop area; Step S2: Send the safety data acquired in real time into the real-time monitoring model one by one for calculation, verify the real-time monitoring results according to the prediction results, and obtain the real-time monitoring results; optimize the real-time monitoring model and the prevention model through joint training, so that the results of the models can be verified between the models, thereby improving the recognition accuracy of the models; Step S3: sending the acquired time series safety data to the prevention model for calculation, optimizing the prediction results according to the real-time monitoring results to obtain the prediction results, matching the prediction results to obtain the prediction control results; performing emergency treatment for the emergency situations in both the real-time monitoring results and the prediction control results; Save the processing results of safety data and generate safety training reports; Step S4: divide the real-time monitoring results and the predictive control results according to the data collection area and the safety category; match the regional control strategy and the regional early warning strategy in the regional control and early warning strategy library according to the real-time monitoring results and the predictive control results in the same area, and automatically control and warn the workshop; Step S5: matching the associated control strategy and the associated warning strategy in the associated control warning strategy library according to the real-time monitoring results and the predictive control results of the same safety category, and performing automatic control and warning; Step S6: Record the safety data and processing results in the workshop. The records will be used to optimize the real-time monitoring model, prevention model, regional control warning strategy library and associated control warning strategy library; the recorded safety data includes original safety data, real-time monitoring results, predictive control results, control strategy and warning strategy. Incorrect processing of abnormal situations in the saved records will be marked, and records exceeding the rated number of times will be added to the regional control warning strategy library and the associated control warning strategy library. The correct records are used as training data for the real-time monitoring model and the prevention model to optimize the model.

2. A safety monitoring method for a paint production workshop according to claim 1, It is characterized in that The pre-trained recognition models include: Independent recognition model training is performed for each source of sound data and image data. Each recognition model is only used for the recognition of corresponding data. The recognition model is built on the basis of support vector machine and is obtained through training and optimization with several annotated historical data. The final output of the pre-trained recognition model is the monitoring result of sound data and image data.

3. A safety monitoring method for a paint production workshop according to claim 1, It is characterized in that Real-time monitoring models and prevention models include: The real-time monitoring model includes: The sensor data processing layer is used to process sensor data, mainly gas concentration sensor data, temperature sensor data and humidity sensor data. The data of each sensor is firstly judged by threshold value to obtain the abnormal weight α, and then the abnormal discrimination value P is obtained according to the change value of the sensor data. The abnormal value K of the sensor is calculated by the abnormal discrimination value and the abnormal weight, and the sensor data processing result is obtained according to the interval where the abnormal value is located. The calculation formula of the abnormal value is: Where α 1 Represents the abnormal weight of the data at the first moment, P 1 Indicates the sensor change value at the first moment; The real-time monitoring layer is used to classify the sensor data processing results, sound monitoring results and image monitoring results to obtain preliminary real-time monitoring results; The result optimization layer is used to optimize the preliminary real-time monitoring results according to the prediction results to obtain the real-time monitoring results; The prevention model includes: The data merging layer is used to reconstruct the sensor data processing results, sound monitoring results and image monitoring results to obtain time-series safety prediction data; The security prediction layer is used to calculate based on the security prediction data and obtain preliminary prediction results; The result optimization layer is used to optimize the preliminary prediction results according to the real-time monitoring results to obtain the prediction results.

4. A safety monitoring method for a paint production workshop according to claim 1, It is characterized in that The joint training of the real-time monitoring model and the prevention model includes: Construct an initial real-time monitoring model and an initial prevention model. The initial real-time monitoring model is constructed based on a deep convolutional neural network, and the initial prevention model is constructed based on a long-short time neural network. The initial real-time monitoring model contains 1 input layer, 1 data calculation layer, 16 hidden layers, and 2 output layers. The initial prevention model contains 1 input layer, 36 hidden layers, and 2 output layers. The last 4 hidden layers of the two models are shared to optimize the results. Obtain some safety data, and annotate real-time abnormal results and prediction results, form the annotated data into a safety data set, and divide the safety data set into a verification set, a test set, and a training set; A certain proportion of data from the safety data training set and validation set is selected to train the initial real-time monitoring model and the initial prevention model, respectively, with real-time abnormal results and prediction results as the targets, to obtain the optimized real-time monitoring model and the optimized prevention model; A certain proportion of data from the security data test set is selected to train the optimized real-time monitoring model and the optimized prevention model. The correct ratio of the output of the final output layer to the actual annotation is taken as the accurate value. The hyperparameters of the two models are adjusted according to the accurate value to obtain the real-time monitoring model and prevention model with the final accuracy rate meeting the standard.

5. A safety monitoring method for a paint production workshop according to claim 1, It is characterized in that The regional control warning strategy library and the associated control warning strategy library include: The regional control and early warning strategy library stores the specific control and early warning strategies for the real-time monitoring results and preventive control results in the region; the control and early warning strategies in the regional control and early warning strategy library are more targeted and can control and warn abnormal situations in the current region in the first place; The associated control and early warning strategy library stores the specific control and early warning strategies for all real-time monitoring results and preventive control results of the same security category. Compared with the regional control and early warning strategy library, the associated control and early warning strategy library associates the security data analysis results between different regions and obtains potential abnormal situation handling strategies through the association between regions.

6. A safety monitoring method for a paint production workshop according to claim 1, It is characterized in that Verification of real-time monitoring results based on prediction results includes: After the real-time monitoring model outputs the results, it is verified through the correspondence between the prediction results and the real-time monitoring results in the real-time prediction association table. The real-time prediction association table stores the direct and indirect correspondence between the prediction results and the real-time monitoring results. The direct correspondence indicates that there is a strong correlation between the prediction results and the real-time monitoring results, and the indirect correlation indicates that there is a weak correlation between the prediction results and the real-time monitoring results. If the verification result is a strong correlation, the verification is passed. If the verification result is a weak correlation, multiple consecutive verifications of weak correlation are required to pass the verification. If the frequency of the verification result that there is no correlation exceeds the threshold, it will be recorded, and the record will be used to update the real-time prediction association table.

7. A safety monitoring method for a paint production workshop according to claim 1, It is characterized in that Generate a safety training report including: Acquire and process the security data processing results in different time periods, including sensor data, sound data, image data and their corresponding anomaly detection results and emergency processing records; perform cluster analysis on the security data, group and classify similar security events according to the clustering results, and generate clustering results for each type of event; perform keyword extraction on the key features in the clustering results, perform statistical analysis based on the frequency of occurrence of the keywords, and generate report content related to various types of safety hazards; predict possible future security risks based on historical data and the temporal characteristics of security events through time series analysis methods, and incorporate the prediction results into the safety training report; the safety training report includes a graphical display of security data, intuitively presenting the security risks and data distribution in different areas.

8. A safety monitoring system for a paint production workshop, It is characterized in that The system applies a safety monitoring method for a paint shop according to any one of claims 1 to 7, and comprises the following modules: The data acquisition module includes a data acquisition unit and a data processing unit. The data acquisition unit is used to obtain sensor data, sound data and image data from various areas of the paint shop. The data processing unit is used to send the sound data and image data to a pre-trained recognition model for abnormality recognition, obtain sound monitoring results and image monitoring results, use the sensor data, sound monitoring results and image monitoring results as safety data, pre-process the safety data, and add safety monitoring tags to the safety data. The real-time monitoring module includes a real-time monitoring unit and a model optimization unit. The real-time monitoring unit contains a real-time monitoring model for calculating the safety data obtained in real time to obtain real-time monitoring results; the model optimization unit is used to optimize the real-time monitoring model and the prevention model, and associate the real-time monitoring model with the prevention model through joint training and record optimization, and use the training set of the record optimization model; The prevention module includes a prediction unit, a prediction control unit, an emergency processing unit and a safety training unit. The prediction unit is used to send the acquired safety data into the prevention model for calculation, and optimize the prediction results according to the real-time monitoring results to obtain the prediction results; the prediction control unit is used to match the prediction control results according to the prediction results; the emergency processing unit is used to handle the emergency situations in the real-time monitoring results and the prevention control results; The safety training unit is used to save the processing results of safety data and generate safety training reports; The early warning module and the control module are integrated in the control early warning module. In addition to the early warning module and the control module, the control early warning module also includes a result division unit and a matching unit. The result division unit is used to divide the real-time monitoring results and the prediction control results according to the data collection area and the safety category; the matching unit is used to match the regional control strategy and the regional early warning strategy in the regional control early warning strategy library according to the real-time monitoring results and the prediction control results in the same area, and match the associated control strategy and the associated early warning strategy in the associated control early warning strategy library according to the real-time monitoring results and the prediction control results in the same safety category; A control module, used for automatically controlling the workshop according to the matched regional control strategy and the associated control strategy; The early warning module is used to automatically warn the workshop according to the matched regional early warning strategy and associated early warning strategy; The recording module is used to record the safety data and processing results in the workshop. The records will be used to optimize the real-time monitoring model, prevention model, regional control and early warning strategy library and associated control and early warning strategy library.

Citation Information

Patent Citations

  • Workshop production safety early warning system and early warning method

    CN115909645A

  • Workshop worker detection system carrying cloud server

    CN116977922A

  • Fire safety monitoring system and method based on alarm area model

    CN117095506A

  • Electric power operation safety monitoring method and system based on edge AI

    CN117235443A

  • Road construction safety monitoring method and system

    CN117575332A