Safety management monitoring system and monitoring method for cigarette factory

Abstract

The invention discloses a safety management monitoring system and monitoring method for a cigarette factory, and belongs to the technical field of safety management. The monitoring system comprises a data acquisition module, a back-end service system and a man-machine interaction platform which are in communication connection with one another and perform data interaction, the data acquisition module is used for acquiring safety and production data of each area of the cigarette factory and sending the safety and production data to the back-end service system for processing and analysis; the back-end service system is used for carrying out risk assessment on the received safety and production data through a deep learning model, generating early warning information according to a corresponding risk assessment result, and sending the early warning information to the human-computer interaction platform; and the man-machine interaction platform is used for receiving the early warning information sent by the back-end service system and reminding a user of risk relief in a preset mode. According to the invention, through a real-time monitoring and early warning mechanism, the safety management level of the cigarette factory is obviously improved, and the occurrence rate of safety accidents is effectively reduced.

Description

Technical Field

[0001] The invention relates to a safety management monitoring system and a monitoring method for a cigarette factory, belonging to the technical field of safety management. Background Art

[0002] Cigarette factories, as an important cornerstone in the traditional manufacturing industry, are facing unprecedented pressure for transformation and upgrading. In the wave of digital transformation of the global manufacturing industry, cigarette factories must keep up with the pace of the times and comprehensively improve their safety management level through the in-depth application of information technology. This is not only an inherent need for the development of the enterprise itself, but also a key step for the entire industry to move towards high-quality development. However, the current situation is not optimistic. At present, the safety management system of cigarette factories still relies heavily on traditional manual supervision and paper records. The drawbacks of this model are particularly prominent in the real-time, accuracy and efficiency of information processing.

[0003] Specifically, traditional safety management and monitoring systems have obvious limitations in monitoring scope, and can often only cover part of the production site. Faced with massive amounts of safety data, the system can often only perform simple surface statistics and preliminary analysis, and the system lacks the ability to conduct deep mining and intelligent prediction, and cannot provide strong data support for decision makers, resulting in safety management decisions often being based on experience rather than data, limiting the improvement of safety management levels. Most systems still remain at the level of passive response, lacking the ability to learn autonomously and adapt to complex safety scenarios, which results in the system often being inaccurate in risk prediction and emergency response, making it difficult to discover and resolve safety hazards in the first place, thus posing huge risks to the company's safe production. Existing emergency response processes mostly rely on fixed patterns and steps, lacking the flexibility to quickly adjust and optimize according to the characteristics of specific safety incidents, resulting in inefficient emergency response. In addition, the user interface design of many safety management systems does not fully consider operational convenience and intuitiveness, resulting in a poor user experience and affecting the system's usability and popularity.

[0004] Therefore, it is urgent to develop a highly information-based and intelligent security management and monitoring platform to specifically solve the relevant technical problems faced in the existing technology. Summary of the invention

[0005] The purpose of the present invention is to provide a safety management monitoring system and monitoring method for a cigarette factory, which aims to collect, process and analyze the safety and production data of various areas of the cigarette factory in real time in an intelligent manner, promptly discover potential safety risks, and remind users to eliminate risks through a human-computer interaction platform, thereby ensuring the safe production of the cigarette factory.

[0006] In order to achieve the above objectives / solve the above technical problems, the present invention is implemented by adopting the following technical solutions.

[0007] On the one hand, the present invention provides a safety management and monitoring system for a cigarette factory, which includes: a data acquisition module, a back-end service system and a human-computer interaction platform that are connected to each other for communication and data exchange; The data collection module is used to collect safety and production data of various areas of the cigarette factory and send the safety and production data to the back-end service system for processing and analysis; The backend service system is used to perform risk assessment on the received safety and production data through a deep learning model, generate warning information according to the corresponding risk assessment results, and send the warning information to the human-computer interaction platform; The human-computer interaction platform is used to receive the warning information sent by the back-end service system and remind the user to eliminate the risk in a preset manner.

[0008] The data acquisition module, the back-end service system and the human-computer interaction platform in the safety management and monitoring system for cigarette factories provided by the present invention establish a communication connection, thereby realizing the interactive flow of data. Among them, the main responsibility of the data acquisition module is to collect the safety and production data of various areas of the cigarette factory, and then pass these data to the back-end service system for in-depth processing and analysis. The back-end service system uses advanced deep learning models to conduct a comprehensive risk assessment of the received safety and production data, and generates corresponding early warning information based on the assessment results. These early warning information will be sent to the human-computer interaction platform in a timely manner. The human-computer interaction platform is responsible for receiving these early warning information, and through preset methods, such as sound alarms, visual prompts, etc., timely remind users to pay attention and take corresponding measures to eliminate risks.

[0009] Optionally, the safety and production data include: environmental data, equipment status data and employee behavior data; The environmental data is collected through sensors deployed at the production site and underground pipe gallery, including: temperature and humidity, smoke concentration and harmful gas concentration; The equipment status data is collected through the monitoring system built into the production equipment, including: vibration frequency, temperature and pressure; The employee behavior data is collected through a video surveillance system deployed at the production site and underground pipe gallery, including employee images or video information.

[0010] Optionally, the backend service system includes: a data processing layer, a risk assessment layer and an early warning notification layer; The data processing layer is used to perform preprocessing operations such as cleaning, de-noising, standardization and filling missing values ​​on the safety and production data, and transmit the preprocessed safety and production data to the risk assessment layer; The risk assessment layer is used to perform image recognition on the pre-processed employee behavior data through a convolutional neural network, and to perform time series prediction on the pre-processed environmental data or equipment status data through a long short-term memory network, and transmit the risk assessment results corresponding to each type of data to the early warning notification layer; The warning notification layer is used to generate warning information according to the risk assessment results and preset thresholds corresponding to various types of data, and transmit the warning information to the human-computer interaction platform.

[0011] Optionally, the preset threshold is obtained by automatically calculating the mean and variance of historical data through an N-sigma algorithm, and the confirmation process includes: Obtain historical data of the environment data, equipment status data, and employee behavior data corresponding to the past three days; Analyze the historical data to obtain normal distribution characteristics corresponding to each data indicator; Based on the normal distribution characteristics, a corresponding three times standard deviation is set as a preset threshold corresponding to each data indicator.

[0012] Optionally, the human-computer interaction platform adopts a role-based access control mechanism to allocate access rights according to different levels of user roles, so that only authorized users can access relevant data stored and processed by the back-end service system and execute risk relief operations.

[0013] In a second aspect, the present invention further provides a cigarette factory safety management and monitoring method applicable to the cigarette factory safety management and monitoring system of the first aspect, comprising: Obtain environmental data, equipment status data, and employee behavior data for each area of ​​the cigarette factory; Inputting the environmental data and equipment status data into a pre-trained long short-term memory network for time series prediction to obtain risk assessment results corresponding to the environmental data and equipment status data; Inputting the employee behavior data into a pre-trained convolutional neural network for image recognition to obtain a risk assessment result corresponding to the employee behavior data; Generate early warning information based on the risk assessment results corresponding to the environmental data, equipment status data, and the risk assessment results corresponding to the employee behavior data to achieve safety management monitoring; The warning information is generated hierarchically according to the risk level corresponding to the risk assessment result, including: level one warning, level two warning, level three warning and level four warning.

[0014] The safety management and monitoring method for cigarette factories proposed in the present invention combines two neural network models, LSTM and CNN, and can collect and analyze the environmental data, equipment status data and employee behavior data of the cigarette factory in real time, generate hierarchical early warning information, and realize efficient and accurate safety management and monitoring.

[0015] Optionally, the environmental data includes: temperature and humidity, smoke concentration and harmful gas concentration; the equipment status data includes: vibration frequency, temperature and pressure; the employee behavior data includes employee image or video information.

[0016] Optionally, the security management and monitoring method also includes data preprocessing, including: cleaning, denoising, standardizing and filling missing values ​​of the environmental data, equipment status data and employee behavior data.

[0017] Optionally, the formula for the standardization process is as follows: ; In the formula, represents the standardized data, Represents the original data, represents the data mean, Represents the standard deviation of the data.

[0018] Optionally, the environmental data and equipment status data are input into a pre-trained long short-term memory network to obtain a corresponding risk assessment result, which is expressed as: ; In the formula, represents the output of the LSTM unit at the current time step, represents a nonlinear activation function, represents the weight of the input to the hidden state, represents the weight from hidden state to hidden state, represents the output of the LSTM unit at the previous time step, represents the first offset, represents the cell state at the current time step, represents the forget gate, represents the cell state at the previous time step, represents the input gate, represents the hyperbolic tangent function, represents the input data of the current time step, represents the second offset, represents the output gate, represents the weight matrix.

[0019] Optionally, a role-based access control mechanism is adopted to assign access rights according to different levels of user roles, so that only authorized users can access the environmental data, equipment status data, employee behavior data, risk assessment results and warning information.

[0020] Compared with the prior art, the present invention has the following beneficial effects: The present invention realizes comprehensive real-time monitoring of key areas of the cigarette factory, including production sites and underground pipe galleries, through multi-dimensional monitoring technology, effectively improving the breadth and depth of monitoring; the present invention also adopts advanced artificial intelligence and machine learning algorithms to conduct in-depth analysis of the collected safety data, provide intelligent prediction and risk assessment, and enhance the decision-making support capabilities; through the ability of autonomous learning and adapting to complex security scenarios, the present invention can also more accurately predict risks and make emergency responses, thereby improving the intelligence level of safety management; by designing a flexible emergency response process, the present invention can also quickly adjust and optimize according to the characteristics of different safety incidents to improve the efficiency and adaptability of emergency response; in addition, the user interface designed by the present invention takes into account the convenience and intuitiveness of operation, which not only improves the user experience, but also effectively enhances the ease of use and popularity of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic flow chart of an embodiment of the safety management and monitoring method for a cigarette factory of the present invention; Figure 2 Shown is a schematic structural diagram of the safety management and monitoring system for a cigarette factory according to the present invention. DETAILED DESCRIPTION

[0022] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. The embodiments of the present invention and the technical features in the embodiments may be combined with each other unless there is a conflict.

[0023] The term "and / or" is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " generally indicates that the associated objects are in an "or" relationship.

[0024] Example 1 This embodiment introduces a safety management and monitoring system for a cigarette factory, which includes: a data acquisition module, a back-end service system and a human-computer interaction platform that are connected to each other for communication and data exchange; The data collection module is used to collect safety and production data of various areas of the cigarette factory and send the safety and production data to the back-end service system for processing and analysis; The backend service system is used to perform risk assessment on the received safety and production data through a deep learning model, generate warning information according to the corresponding risk assessment results, and send the warning information to the human-computer interaction platform; The human-computer interaction platform is used to receive the warning information sent by the back-end service system and remind the user to eliminate the risk in a preset manner.

[0025] In the actual application process of this embodiment, the data acquisition module, as the front end of the system, is first responsible for extensively collecting safety and production data from various areas of the cigarette factory. These data include but are not limited to key information such as temperature, humidity, smoke concentration, and equipment operation status. The collected data will be sent to the back-end service system in real time for further processing and analysis. Then, the back-end service system receives data from the data acquisition module and uses advanced deep learning models to conduct in-depth risk assessment on these data. Through intelligent algorithms, the back-end service system can accurately identify potential safety hazards and generate corresponding early warning information based on the risk assessment results. Then, the human-computer interaction platform conveys the early warning information generated by the back-end service system to the user. The platform can not only display the early warning information in an intuitive way, but also ensure that users can promptly notice the risks and take corresponding measures to eliminate them through preset sound alarms, visual prompts and other reminders to reduce the probability of safety accidents. In general, this embodiment enables the monitoring system to effectively improve the safety management level of the cigarette factory through real-time monitoring and intelligent early warning, and provides a strong guarantee for the safe production of the cigarette factory.

[0026] Example 2 refer to Figure 2 Based on Example 1, this example realizes a comprehensive upgrade of the safety management of the cigarette factory through advanced Web technology and modular design concepts. The following are the key technical features and implementation details of the system.

[0027] 1. Data acquisition module As the front end of the monitoring system, the data acquisition module is responsible for collecting various safety and production data in the cigarette factory in real time and accurately, among which the safety and production data include: environmental data, equipment status data and employee behavior data.

[0028] Specifically, the environmental data is collected through sensors installed at the production site and underground pipe corridors, including: temperature and humidity, smoke concentration and harmful gas concentration; the equipment status data is collected through the built-in monitoring system of the production equipment, including: vibration frequency, temperature and pressure; the employee behavior data is collected through the video monitoring system installed at the production site and underground pipe corridors, including employee images or video information, such as whether they are wearing a safety helmet.

[0029] 2. Backend Service Architecture The backend service developed in Java is designed to build an efficient, stable and reliable system core. As the core of the monitoring system, the backend service system is not only responsible for receiving, processing and analyzing data from the data acquisition module, but also responsible for processing business logic, database interaction and external system integration. Through the RESTful API design principle, a data access interface with front-end and back-end separation is provided.

[0030] Specifically, the system adopts a modular design, clearly divides the business logic, and optimizes the data processing process through middleware such as message queues and cache services to improve the overall response speed. In terms of database interaction, connection pool management and query optimization are implemented to ensure the security and efficiency of data access. At the same time, through the API gateway and microservice architecture, the integration capability with external systems is enhanced to achieve flexible service-to-service communication. In the design of RESTful API, the standardization principle is followed, and version control and automatic document generation are introduced to improve the efficiency of development and maintenance. In terms of security, an authentication and authorization mechanism based on OAuth 2.0 is implemented, and end-to-end encryption technology is used to protect data security. A comprehensive back-end testing strategy, including integration testing, performance testing, and automated testing, ensures the stability and efficiency of the service.

[0031] The back-end service system at least includes: a data processing layer, a risk assessment layer and an early warning notification layer.

[0032] The data processing layer is used to perform preprocessing operations on the safety and production data, such as cleaning, de-noising, standardization, and filling missing values, and transmit the preprocessed safety and production data to the risk assessment layer.

[0033] Specifically, the data is cleaned, denoised, standardized, and missing values ​​are filled to ensure data quality and consistency. The formula for the standardization process is as follows: ; In the formula, represents the standardized data, Represents the original data, represents the data mean, Represents the standard deviation of the data.

[0034] The risk assessment layer is used to perform image recognition on the preprocessed employee behavior data through a convolutional neural network, and to perform time series prediction on the preprocessed environmental data or equipment status data through a long short-term memory network, and transmit the risk assessment results corresponding to each type of data to the early warning notification layer.

[0035] Specifically, the risk assessment layer is a risk assessment engine that integrates artificial intelligence and machine learning algorithms. It is the core of the monitoring system and has powerful data processing and analysis capabilities. The engine integrates real-time data collected by sensors in various areas of the cigarette factory and conducts in-depth data preprocessing and feature engineering to ensure the quality and accuracy of input data. Using a hybrid model of long short-term memory networks (LSTM) and convolutional neural networks (CNN), the engine is able to capture long-term dependencies in time series and complex patterns in spatial data, thereby achieving accurate identification and quantitative assessment of potential safety risks. The engine's adaptive learning capabilities enable it to continuously optimize the model based on new data and feedback to improve the accuracy of predictions.

[0036] The following is the specific implementation process of risk assessment: First, the preprocessed data is input into the trained machine learning model. The model uses a hybrid model of long short-term memory network (LSTM) and convolutional neural network (CNN) to capture the long-term dependencies in time series and the complex patterns of spatial data. Among them, the environmental data and equipment status data are input into the pre-trained long short-term memory network to obtain the corresponding risk assessment results, which are expressed as: ; In the formula, represents the output of the LSTM unit at the current time step, which can be regarded as the memory or state of the network at that time step. represents a nonlinear activation function, and a sigmoid function is usually used. In this embodiment, 0.5 is taken to introduce nonlinear characteristics and help the model learn complex patterns. represents the weight of the input to the hidden state, represents the weight from hidden state to hidden state, represents the output of the LSTM unit at the previous time step, Represents the first bias, which is used to adjust the output of the activation function to help the model better learn data features. represents the cell state at the current time step, which carries information for calculating the next time step, represents the forget gate, represents the cell state at the previous time step, represents the input gate, Represents the hyperbolic tangent function, which is used as an activation function for the calculation of candidate memory cells. Its value range is (-1, 1), which helps to compress the value into this range for subsequent calculations. represents the input data of the current time step, represents the second offset, represents the output gate, Represents the weight matrix, which is used to map the input data of the current time step and the hidden state of the previous time step to the calculation of the candidate memory cell. Among them, the forget gate, input gate, and output gate control the flow of information, determining how much past information is retained, how much new information is added, and how much information is output to the next hidden state.

[0037] In actual application, the model identifies potential safety risks based on the input data and predicts possible safety incidents. For example, if the concentration of harmful gases is detected to exceed the threshold, the model will assess the risk of gas leakage and predict the scope and severity of the possible impact.

[0038] The warning notification layer is used to generate warning information according to the risk assessment results and preset thresholds corresponding to various types of data, and transmit the warning information to the human-computer interaction platform.

[0039] Specifically, the model outputs the risk assessment results, and the system dynamically generates warning information based on the risk assessment results, such as environmental parameter exceeding the standard, abnormal equipment status, and unsafe employee behavior, and promptly notifies relevant personnel through the user interface. According to the risk level, the system will propose corresponding response strategies. For example, for high-risk warnings, the system may recommend immediate evacuation of personnel and initiation of emergency response procedures; for medium and low risks, it may recommend regular inspections and maintenance.

[0040] Specifically, early warning information is generated hierarchically based on risk assessment results: Warning Level 1 (P1): Indicates the most serious risk and requires immediate action, which may involve evacuation of personnel or emergency shutdown.

[0041] Level 2 Warning (P2): Indicates a higher risk, requiring relevant departments to closely monitor and prepare emergency measures.

[0042] Level 3 Warning (P3): indicates medium risk and requires attention and regular inspection.

[0043] Level 4 Warning (P4): Indicates the lowest risk and is only a reminder. No immediate action is required.

[0044] The above process automatically calculates the mean and variance of the historical data of each indicator through the N-sigma algorithm, and sets the relevant thresholds corresponding to each type of data accordingly. Specifically, the historical 3-day data can be used to analyze the normal distribution of the indicator and set 3 times the standard deviation as the threshold to ensure that the probability of the indicator deviating from the expectation is very small.

[0045] In addition, this embodiment also provides a dynamic threshold function, which adaptively tracks the indicator change trend according to the indicator focus direction and continuous cycle configuration. This design ensures that when the potential risk reaches a critical level, early warnings can be issued to relevant personnel in a timely manner. The generation of early warning information is dynamic, which means that they will be adjusted according to the continuous monitoring and analysis results of the risk assessment engine to provide the most accurate and timely security information.

[0046] 3. Front-end human-computer interaction platform First of all, it is worth mentioning that in order to further enhance the security and data protection of the system, a role-based access control (RBAC) mechanism is also adopted. RBAC is a security strategy that assigns access rights based on the user's role, ensuring that only authorized users can access sensitive data and perform specific operations. This access control mechanism improves data security and also provides customized access rights for different levels of users to meet the needs and responsibilities of different users.

[0047] In addition, the RBAC mechanism also includes detailed auditing and logging of user operations, which helps to track and review the user's access and operation history to sensitive data, further enhancing the transparency and traceability of the system. Through this meticulous access control, it ensures that the system maintains the integrity and security of the data while providing the necessary flexibility and functionality.

[0048] In the implementation of the front-end user interface, HTML5 and JavaScript are used to build a cross-platform compatible user interface to ensure that the system can run seamlessly on different devices and browsers. Bulma and CSS frameworks are used to implement responsive layout design, so that the interface can adapt to different screen sizes and provide a consistent user experience. Ajax technology is used to achieve asynchronous data exchange with the back end to improve the interactivity and response speed of the user interface.

[0049] Specifically, CSS media queries are used to optimize for different screen sizes to ensure the best view on mobile phones, tablets, desktops, etc. Dynamic layout adjustment function is implemented to automatically adjust the interface layout according to the user's screen size and orientation changes.

[0050] Secondly, the server push is implemented using Ajax long polling technology. When there is new data or status update, the server actively sends information to the client. The WebSocket protocol is integrated to provide full-duplex communication capabilities for functions that require real-time communication, reducing the delay caused by page refresh. In addition, CSS, JavaScript and image resources are compressed to reduce loading time and bandwidth consumption, unit tests are written for front-end code to ensure the correctness and stability of functions, end-to-end testing is implemented, user operation processes are simulated, and the overall function of the system is verified. Lazy loading technology is used to load interface elements and data on demand to improve the loading speed of the first screen.

[0051] 4. Modular system design The monitoring system adopts advanced modular design concepts to ensure the independence of functions and efficient interaction between modules, including but not limited to staffing, workshop notifications, meeting minutes, work arrangements, safety monitoring, etc. The modular design makes the system easy to expand and maintain, while allowing customized development according to the specific needs of different cigarette factories.

[0052] In the safety management and monitoring system, the relevant modules in the following human-computer interaction platform cover multiple aspects of on-site safety management, as described below.

[0053] New form module: allows users to customize and submit forms related to security management to achieve standardized data entry and process management.

[0054] Keyword search module: provides an efficient search mechanism that enables users to quickly retrieve required data through key information and optimize the information search process.

[0055] Information export module: supports exporting security management data in multiple formats for further analysis and record keeping.

[0056] Underground Pipe Gallery Management Module: Focuses on the special safety requirements of underground pipe gallery, including: Off-site QR code information entry module: Use QR code technology to achieve rapid data entry and verification, improving the efficiency and accuracy of safety management in underground utility tunnel areas.

[0057] The advantage of modular design is its high scalability and maintainability. As the subsequent safety management needs of cigarette factories continue to change and technology continues to advance, the system can easily introduce new modules or update existing modules without affecting the stability of other parts. In addition, the modular architecture also supports customized development for the specific needs of different cigarette factories, allowing the system to better adapt to the business processes and safety standards of different users.

[0058] Example 3 refer to Figure 1 Based on the same inventive concept as in Example 1 or 2, this embodiment introduces a safety management and monitoring method for a cigarette factory, which includes: Obtain environmental data, equipment status data, and employee behavior data for each area of ​​the cigarette factory; Inputting the environmental data and equipment status data into a pre-trained long short-term memory network for time series prediction to obtain risk assessment results corresponding to the environmental data and equipment status data; Inputting the employee behavior data into a pre-trained convolutional neural network for image recognition to obtain a risk assessment result corresponding to the employee behavior data; Generate early warning information based on the risk assessment results corresponding to the environmental data, equipment status data, and the risk assessment results corresponding to the employee behavior data to achieve safety management monitoring; Among them, the warning information is generated in levels according to the risk degree corresponding to the risk assessment result, including: level one warning, level two warning, level three warning and level four warning.

[0059] It is worth noting that the above risk assessment results include the following forms: Risk score: A value of 0-10 points, indicating the level of risk (low risk output value is 0-4, medium risk output value is 5-7, high risk output value is 8-10). Classification results: Directly evaluate the risk into different categories, such as "safe" or "unsafe", "normal" or "abnormal", etc. Warning signal: When the risk reaches the threshold of 6 points, the system will issue a warning signal, indicating that action needs to be taken. In addition, for employee behavior data, CNN will output analysis results on employee behavior, such as whether safety regulations are followed and whether there are potential safety risks.

[0060] In the actual application process of this embodiment, the acquired environmental data and equipment status data are input into a pre-trained long short-term memory network (LSTM), which can process time series data and capture long-term dependencies in the data, thereby performing risk assessment on the environmental data and equipment status data. The acquired employee behavior data is input into a pre-trained convolutional neural network (CNN), which can extract features from image or video data, analyze employee behavior, and determine whether employee behavior has potential risks. Then, the above LSTM network and CNN network process the input data respectively and output their respective risk assessment results. Then, according to the above risk assessment results, corresponding warning information is generated, and different levels of warning information correspond to different risk levels and response measures. Finally, the generated warning information is sent to the human-computer interaction platform, and relevant managers can take corresponding risk management measures according to the level and content of the warning information, such as strengthening inspections and launching emergency plans.

[0061] In general, this embodiment combines the long short-term memory network (LSTM) and the convolutional neural network (CNN) to analyze the environmental data, equipment status data and employee behavior data of the cigarette factory, generate early warning information, and achieve effective and timely safety management and monitoring.

[0062] Example 4 Based on Example 3, this example will introduce the specific implementation process of safety management monitoring for cigarette factories, and the relevant steps are described as follows.

[0063] 1. Data Acquisition and Preprocessing Collect various safety and production data in the cigarette factory, including environmental data, equipment status data and employee behavior data. The environmental data is collected through sensors deployed at the production site and underground pipe gallery, including temperature and humidity, smoke concentration and harmful gas concentration; the equipment status data is collected through the built-in monitoring system of the production equipment, including vibration frequency, temperature and pressure; the employee behavior data is collected through the video monitoring system deployed at the production site and underground pipe gallery, including employee images or video information, such as whether they are wearing a helmet.

[0064] Furthermore, the data obtained above is preprocessed. Specifically, it includes: cleaning, denoising, standardizing and filling missing values ​​of the environmental data, equipment status data and employee behavior data. The formula of the standardization process is as follows: ; In the formula, represents the standardized data, Represents the original data, represents the data mean, Represents the standard deviation of the data.

[0065] II. Regional Risk Assessment The preprocessed environmental data and equipment status data are input into the pre-trained long short-term memory network, and the preprocessed employee behavior data are input into the pre-trained convolutional neural network to obtain the risk assessment results corresponding to each type of data.

[0066] When discussing environmental data, equipment status data, and employee behavior data, convolutional neural networks (CNN) and long short-term memory networks (LSTM) are two powerful deep learning models, each with unique application scenarios and advantages.

[0067] Convolutional neural network is a type of feedforward neural network with deep structure and convolution calculation. It is one of the representative algorithms of deep learning. It has the ability of representation learning and can classify the input information in a translation-invariant manner according to its hierarchical structure. Therefore, it is also called "translation-invariant artificial neural network". CNN has significant advantages in image recognition, target detection and other fields, thanks to its special layer structure and working mechanism: Input layer: accepts raw data as input. For image processing tasks, the input layer is usually a picture, which is specifically represented by a multi-dimensional matrix. Convolution layer: extracts local features in the image through convolution operation. The convolution kernel (a two-dimensional matrix less than or equal to the size of the input image) slides on the input image, calculates the dot product between the local area and the convolution kernel, and generates a feature map. Pooling layer: also called downsampling layer, is used to extract the most representative features in the feature map, helps reduce unnecessary features, thereby reducing the risk of overfitting and reducing the dimension of the data. Fully connected layer: integrates features and completes the final prediction. All neurons are connected to all neurons in the previous layer, and the feature vector output by the pooling layer is processed to integrate the features of the image into a higher-level feature representation. Output layer: outputs the final prediction result.

[0068] In actual applications, CNN can effectively extract and identify features of images or video information in environmental data and equipment status data (such as images captured by surveillance cameras, equipment appearance images, etc.). For example, in environmental monitoring, CNN can identify whether employees are wearing helmets in images.

[0069] Long short-term memory network is a special recurrent neural network (RNN) that can better process sequence data and can remember for a long time. LSTM solves the long-term dependency problem of traditional RNN by introducing a gate mechanism to control the entry, retention and exit of information. The core components of LSTM include input gate, forget gate, output gate and cell unit. Among them, the input gate is used to control the entry of input information into the cell unit; the forget gate is used to control the retention of the hidden state; the output gate is used to control the output of the hidden state; the cell unit is used to store and update the hidden state.

[0070] LSTM can effectively model and predict the time series information in employee behavior data (such as employee operation records, behavior trajectories, etc.). For example, in employee behavior analysis, LSTM can identify employee behavior patterns and predict employee future behavior trends. In addition, LSTM can also be used to process time series information in environmental data and equipment status data, such as air quality change trends in environmental monitoring and equipment performance degradation trends in equipment status monitoring.

[0071] Specifically, the preprocessed environmental data and equipment status data are input into the pre-trained long short-term memory network to obtain the corresponding risk assessment results, which are expressed as: ; In the formula, represents the output of the LSTM unit at the current time step, which can be regarded as the memory or state of the network at that time step. represents a nonlinear activation function, and a sigmoid function is usually used. In this embodiment, 0.5 is taken to introduce nonlinear characteristics and help the model learn complex patterns. represents the weight of the input to the hidden state, represents the weight from hidden state to hidden state, represents the output of the LSTM unit at the previous time step, Represents the first bias, which is used to adjust the output of the activation function to help the model better learn data features. represents the cell state at the current time step, which carries information for calculating the next time step, represents the forget gate, represents the cell state at the previous time step, represents the input gate, Represents the hyperbolic tangent function, which is used as an activation function for the calculation of candidate memory cells. Its value range is (-1, 1), which helps to compress the value into this range for subsequent calculations. express, represents the input data of the current time step, represents the second offset, represents the output gate, Represents the weight matrix, which is used to map the input data of the current time step and the hidden state of the previous time step to the calculation of the candidate memory cell. Among them, the forget gate, input gate, and output gate control the flow of information, determining how much past information is retained, how much new information is added, and how much information is output to the next hidden state.

[0072] These parameters work together on the LSTM unit, enabling it to process sequence data and capture long-term dependencies. Each parameter plays a specific role and together implements the gating mechanism and memory function of LSTM.

[0073] 3. Dynamic warning generation The model outputs the risk assessment results, and the system dynamically generates warning information based on the risk assessment results, such as environmental parameter exceeding the standard, abnormal equipment status, and unsafe employee behavior, and promptly notifies relevant personnel through the user interface. According to the risk level, the system will propose corresponding response strategies. For example, for high-risk warnings, the system may recommend immediate evacuation of personnel and initiation of emergency response procedures; for medium and low risks, it may recommend regular inspections and maintenance.

[0074] Specifically, early warning information is generated hierarchically based on risk assessment results: Warning Level 1 (P1): Indicates the most serious risk and requires immediate action, which may involve evacuation of personnel or emergency shutdown.

[0075] Level 2 Warning (P2): Indicates a higher risk, requiring relevant departments to closely monitor and prepare emergency measures.

[0076] Level 3 Warning (P3): indicates medium risk and requires attention and regular inspection.

[0077] Level 4 Warning (P4): Indicates the lowest risk and is only a reminder. No immediate action is required.

[0078] It is worth mentioning that in order to further enhance the security and data protection of the system, a role-based access control (RBAC) mechanism is also adopted. RBAC is a security strategy that assigns access rights based on the user's role to ensure that only authorized users can access sensitive data and perform specific operations. This access control mechanism improves data security and also provides customized access rights for different levels of users to meet the needs and responsibilities of different users.

[0079] In addition, the RBAC mechanism also includes detailed auditing and logging of user operations, which helps to track and review the user's access and operation history to sensitive data, further enhancing the transparency and traceability of the system. Through this meticulous access control, it ensures that the system maintains the integrity and security of the data while providing the necessary flexibility and functionality.

[0080] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0082] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0083] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0084] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A safety management and monitoring system for a cigarette factory, characterized in that: include: Data collection modules, back-end service systems and human-computer interaction platforms that communicate with each other and exchange data; The data collection module is used to collect safety and production data of various areas of the cigarette factory and send the safety and production data to the back-end service system for processing and analysis; The backend service system is used to perform risk assessment on the received safety and production data through a deep learning model, generate warning information according to the corresponding risk assessment results, and send the warning information to the human-computer interaction platform; The human-computer interaction platform is used to receive the warning information sent by the back-end service system and remind the user to eliminate the risk in a preset manner.

2. The safety management and monitoring system for cigarette factories according to claim 1 is characterized in that: The safety and production data include: environmental data, equipment status data and employee behavior data; The environmental data is collected through sensors deployed at the production site and underground pipe gallery, including: temperature and humidity, smoke concentration and harmful gas concentration; The equipment status data is collected through the monitoring system built into the production equipment, including: vibration frequency, temperature and pressure; The employee behavior data is collected through a video surveillance system deployed at the production site and underground pipe gallery, including employee images or video information.

3. The safety management and monitoring system for cigarette factories according to claim 2 is characterized in that: The back-end service system includes: a data processing layer, a risk assessment layer and an early warning notification layer; The data processing layer is used to perform preprocessing operations such as cleaning, de-noising, standardization and filling missing values ​​on the safety and production data, and transmit the preprocessed safety and production data to the risk assessment layer; The risk assessment layer is used to perform image recognition on the pre-processed employee behavior data through a convolutional neural network, and to perform time series prediction on the pre-processed environmental data or equipment status data through a long short-term memory network, and transmit the risk assessment results corresponding to each type of data to the early warning notification layer; The warning notification layer is used to generate warning information according to the risk assessment results and preset thresholds corresponding to various types of data, and transmit the warning information to the human-computer interaction platform.

4. The safety management and monitoring system for cigarette factories according to claim 3, characterized in that: The preset threshold is obtained by automatically calculating the mean and variance of historical data through the N-sigma algorithm. The confirmation process includes: Obtain historical data of the environment data, equipment status data, and employee behavior data corresponding to the past three days; Analyze the historical data to obtain normal distribution characteristics corresponding to each data indicator; Based on the normal distribution characteristics, a corresponding three times standard deviation is set as a preset threshold corresponding to each data indicator.

5. The safety management and monitoring system for cigarette factories according to claim 4, characterized in that: The human-computer interaction platform adopts a role-based access control mechanism to allocate access rights according to different levels of user roles, so that only authorized users can access the relevant data stored and processed by the back-end service system and the execution of risk relief operations.

6. A cigarette factory safety management and monitoring method applicable to the cigarette factory safety management and monitoring system according to any one of claims 1 to 5, characterized in that: include: Obtain environmental data, equipment status data, and employee behavior data for each area of ​​the cigarette factory; Inputting the environmental data and equipment status data into a pre-trained long short-term memory network for time series prediction to obtain risk assessment results corresponding to the environmental data and equipment status data; Inputting the employee behavior data into a pre-trained convolutional neural network for image recognition to obtain a risk assessment result corresponding to the employee behavior data; Generate early warning information based on the risk assessment results corresponding to the environmental data, equipment status data, and the risk assessment results corresponding to the employee behavior data to achieve safety management monitoring; Among them, the warning information is generated in levels according to the risk degree corresponding to the risk assessment result, including: level one warning, level two warning, level three warning and level four warning.

7. The security management and monitoring method according to claim 6, characterized in that: The environmental data includes: temperature and humidity, smoke concentration and harmful gas concentration; the equipment status data includes: vibration frequency, temperature and pressure; the employee behavior data includes employee image or video information.

8. The security management and monitoring method according to claim 7, characterized in that: It also includes data preprocessing, including: cleaning, denoising, standardizing and filling missing values ​​of the environmental data, equipment status data and employee behavior data; The formula for the standardization process is as follows: ； In the formula, represents the standardized data, Represents the original data, represents the data mean, Represents the standard deviation of the data.

9. The security management and monitoring method according to claim 8, characterized in that: The environmental data and equipment status data are input into the pre-trained long short-term memory network to obtain the corresponding risk assessment results, which are expressed as: ； In the formula, represents the output of the LSTM unit at the current time step, represents a nonlinear activation function, represents the weight of the input to the hidden state, represents the weight from hidden state to hidden state, represents the output of the LSTM unit at the previous time step, represents the first offset, represents the cell state at the current time step, represents the forget gate, represents the cell state at the previous time step, represents the input gate, represents the hyperbolic tangent function, represents the input data of the current time step, represents the second offset, represents the output gate, represents the weight matrix.

10. The security management and monitoring method according to claim 9, characterized in that: A role-based access control mechanism is adopted to assign access rights according to different levels of user roles, so that only authorized users can access the environmental data, equipment status data, employee behavior data, risk assessment results and warning information.

Images