Smart factory risk management and control method of visual large model

Abstract

The invention discloses a smart factory risk management and control method based on a visual large model, and the method comprises the following steps: S1, collecting data, and carrying out the processing and analysis of the data: collecting factory data, and then carrying out the processing and analysis of the data, and obtaining risk data; s2, constructing a risk model and establishing a risk level: establishing the risk model through a visual inspection technology and a mechanical learning technology according to the risk data, and establishing the risk level according to the condition in the factory, S3, arranging monitoring equipment, and establishing network connection; according to the risk management and control method designed by the invention, the danger source and the risk point in the factory can be directly and quickly found through an image extraction mode, then the risk probability can be quickly obtained through calculation of the danger source and the risk point, information can be quickly sent out through the risk probability, personnel in the factory can timely process the risk, and the risk management and control efficiency is improved. The timeliness and response speed of risk management and control are ensured, and the burden of a factory is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of smart factories, and in particular to a smart factory risk management and control method based on a large visual model. Background Art

[0002] The smart factory is a new stage in the development of modern factory informatization. It uses advanced technologies such as the Internet of Things, big data, and artificial intelligence to help factory production equipment, process flow, production planning, quality control, supply chain and other systems have the characteristics of digitalization, automation, and intelligence, and realize efficient, flexible, and customizable production methods; smart factories use digital means to collect various production data in real time and analyze and apply them to further optimize and control production processes, thereby improving production efficiency and quality, and reducing production costs and risks. Factories use safety supervisors to supervise the behavior of industrial production operators in the enterprise. For example, the safety supervision technology solutions of most chemical processing companies are still manual monitoring and manual safety supervision. Although there are also infrared surveillance cameras for fire sensing monitoring, they are far from meeting the construction requirements of smart factories.

[0003] Such as the authorization announcement number CN117541067B, the authorization announcement date is 20240426, a process intelligent factory risk analysis method, including the following steps: S1: obtain the process parameters to be analyzed of the process intelligent factory; S2: obtain the guide words matched by the process parameters to be analyzed; S3: obtain the consequences caused by the deviation; obtain the direct cause of the deviation; obtain the indirect cause of the deviation based on the direct cause of the deviation; S4: obtain the initial risk caused by the deviation based on the indirect cause of the deviation and the consequences caused by the deviation; obtain the residual risk caused by the deviation based on the safety measures and the initial risk; obtain the risk analysis result based on the residual risk; S5: obtain the next guide word matching the process parameter to be analyzed, repeat S3-S4, until the matching guide words of the process parameter to be analyzed are all analyzed; S6: obtain the next process parameter to be analyzed, repeat S2-S5, until all the process parameters to be analyzed are analyzed. The present invention can comprehensively, comprehensively and accurately analyze the risks of process intelligent factories.

[0004] The above-mentioned and existing analyses of risks within the factory are rather cumbersome and time-consuming, resulting in the inability of workers to control risks in a timely manner when risks occur, causing significant losses to the factory. At the same time, it is far from meeting the construction requirements of a smart factory and will increase the burden on the factory. Therefore, there is an urgent need to design a smart factory risk control method based on a large visual model to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a smart factory risk management method for a large visual model to solve the above-mentioned deficiencies in the prior art.

[0006] In order to achieve the above object, the present invention provides the following technical solutions: A smart factory risk management method based on a visual large model includes the following steps: S1. Collect data, process and analyze the data: collect factory data, then process and analyze the data to obtain risk data; S2. Build a risk model and set risk levels: Based on the risk data, build a risk model through visual inspection technology and machine learning technology, and set risk levels based on the situation in the factory; S3. Arrange monitoring equipment and establish network connection: Select appropriate monitoring equipment, fix it in a suitable location in the factory, and set up a network connection device. The monitoring equipment is connected to the main control server through a network connection device; S4. Collect data information and use models to predict risks: Monitoring equipment collects data information, and risk models identify and extract data, and estimate risks based on the identification and extraction results; S5. Issue risk information and manage risks: When the model predicts risks, an alarm message is issued to remind workers to manage and supervise the risks estimated in the factory.

[0007] The present invention provides a smart factory risk management method based on a visual large model. The risk management method can directly and quickly find the danger sources and risk points in the factory by way of image extraction, and then quickly obtain a risk probability by calculating the danger sources and risk points. Information can be quickly issued through the risk probability, so that factory personnel can deal with the risk in a timely manner, thereby ensuring the timeliness and response speed of risk management and control, and greatly reducing the burden on the factory.

[0008] In another embodiment provided by the present invention, the risk data in step S1 includes hazard sources and risk points in the factory.

[0009] It should be noted that risk points refer to risky parts, facilities, locations and areas, as well as risky operations carried out in specific parts, facilities, locations and areas, or a combination of the above. For example, hazardous chemical tank areas, distillation equipment, loading and unloading operations, and refrigeration equipment are risk points; tank dumping operations in tank areas, hot work operations in fire-proof areas, and the transportation process of hazardous chemicals are also risk points; Hazard source refers to the root cause, state or behavior, or a combination of them, which may cause personal injury and / or health damage and / or property loss.

[0010] Among them: Root cause refers to a physical entity that has energy or generates and releases energy, such as lifting equipment, electrical equipment, pressure vessels, etc. Behavior refers to the decision-making behavior, management behavior, and operation behavior of decision-makers, managers, and practitioners. State refers to the state of objects and the state of the environment, etc.

[0011] When analyzing the factors that cause casualties to people, affect people's physical health, and even lead to diseases during the production process, hazard sources can be called dangerous and harmful factors, which are divided into four categories: "human factors", "object factors", "environmental factors", and "management factors". Human factors refer to the dangerous and harmful factors from personnel themselves or of human nature during production activities; object factors refer to the dangerous and harmful factors existing in machinery, equipment, facilities, materials, etc.; environmental factors refer to the dangerous and harmful factors in the production operation environment; management factors refer to the dangerous and harmful factors caused by the lack of management and management responsibilities.

[0012] In another embodiment provided by the present invention, the risk model in step S2 is composed of a visual module and a pre-judgment module. The visual module is established based on visual detection technology. The visual module is divided into an image feature extraction unit and a text link unit. The image feature extraction unit is used to identify and extract the collected data. The image feature extraction unit extracts and identifies image information through five methods: noise reduction processing, image enhancement, color image to grayscale image conversion, grayscale image to binary image conversion, edge detection / splitting, and histogram matching / contour matching. The text link unit selects appropriate text information to match the extracted image according to the result extracted by the image feature extraction unit. Risk data is stored in the text link unit. The pre-judgment module is established based on machine learning technology. The pre-judgment module pre-judges risks according to the following calculation formula, where the formula is as follows:

[0013] Among them, S is the risk probability. When S≥1, the risk is high and key attention and control are required. When 0<S<1, the risk is relatively low and attention is needed. When S = 0, there is no risk. N is the number of hazard sources and risk points in a certain area within the factory. M is the maximum number of hazard sources and risk points allowed in a certain area within the factory. n is the level of the accident hazard degree when the hazard sources and risk points in a certain area within the factory occur together. m is the maximum allowed accident hazard degree when the hazard sources and risk points in a certain area within the factory occur together. There are three levels of risk in step S2, which are specifically as follows: Level 3 risk: No risk; Level 2 risk: Slightly dangerous, attention is needed; Level 1 risk: Control and rectification are required.

[0014] In another embodiment provided by the present invention, the monitoring device in step S3 is an AI intelligent monitoring camera, the layout of the monitoring device in step S3 should fully cover the key areas of the factory, and the network connection device in step S3 is an industrial computer.

[0015] In another embodiment provided by the present invention, there are two ways of issuing alarm information in step S5, one is to directly issue problem information through speakers or alarms in the factory, and the other is to contact factory personnel through text messages, phone calls, etc. The supervision and processing in step S5 includes four methods: risk avoidance, loss control, risk transfer, and risk retention.

[0016] It should be noted that: ‌Risk avoidance‌: Consciously giving up risky behavior and completely avoiding specific risks of loss. This approach is suitable for situations where people are extremely risk-averse or when there are other target options with lower risks.

[0017] ‌‌Loss Control‌: Make plans and take measures to reduce the possibility of loss or reduce actual loss. Pre-event control is mainly to reduce the probability of loss, while in-event and post-event control is mainly to reduce the actual loss.

[0018] ‌Risk transfer‌: Risk is transferred to other participants or insurance companies through contracts. The main forms include contract transfer and insurance transfer.

[0019] ‌‌Risk retention‌: that is, risk assumption. When facing losses, the investment entity uses the funds it can gather to make up for the losses. Planned self-insurance and unplanned self-retention are the two main forms.

[0020] In the above technical solution, the present invention provides a smart factory risk management method of a large visual model, which has the following beneficial effects: (1) The risk control method designed by the present invention can directly and quickly find the danger sources and risk points in the factory through image extraction. Then, by calculating the danger sources and risk points, a risk probability can be quickly obtained. Information can be quickly issued through the risk probability, allowing factory personnel to deal with risks in a timely manner, thereby ensuring the timeliness and response speed of risk control and greatly reducing the burden on the factory.

[0021] (2) The risk management method designed by the present invention can predict risks and issue timely warnings to areas where risks are about to occur, so that factory personnel can deal with the risks in a timely manner, thereby ensuring the safety of the factory and reducing the losses of the factory. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0023] Figure 1 A method flow chart of an embodiment of a smart factory risk management method for a visual large model of the present invention is provided.

[0024] Figure 2 A schematic diagram of the risk model structure provided for an embodiment of a smart factory risk management and control method for a visual large model of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0026] like Figure 1-2 As shown, a smart factory risk management method of a visual large model provided by an embodiment of the present invention includes the following steps: S1. Collect data, process and analyze the data: collect factory data, then process and analyze the data to obtain risk data; It should be noted that the processing and analysis of data includes data timeliness judgment, data integrity judgment, data classification and summarization processing, and data exclusion processing.

[0027] S2. Build a risk model and set risk levels: Based on the risk data, establish a risk model through visual inspection technology and machine learning technology, and set risk levels based on the situation in the factory: S3. Arrange monitoring equipment and establish network connection: Select appropriate monitoring equipment, fix it in a suitable location in the factory, and set up a network connection device. The monitoring equipment is connected to the main control server through a network connection device; S4. Collect data information and use models to predict risks: Monitoring equipment collects data information, and risk models identify and extract data, and estimate risks based on the identification and extraction results: S5. Issue risk information and manage risks: When the model predicts risks, an alarm message is issued to remind workers to manage and supervise the risks estimated in the factory.

[0028] The present invention provides a smart factory risk management method based on a visual large model. The risk management method can directly and quickly find the danger sources and risk points in the factory by way of image extraction, and then quickly obtain a risk probability by calculating the danger sources and risk points. Information can be quickly issued through the risk probability, so that factory personnel can deal with the risk in a timely manner, thereby ensuring the timeliness and response speed of risk management and control, and greatly reducing the burden on the factory.

[0029] In another embodiment provided by the present invention, the risk data in step S1 includes hazard sources and risk points in the factory.

[0030] It should be noted that risk points refer to risky parts, facilities, locations and areas, as well as risky operations carried out in specific parts, facilities, locations and areas, or a combination of the above. For example, hazardous chemical tank areas, distillation equipment, loading and unloading operations, and refrigeration equipment are risk points; tank dumping operations in tank areas, hot work operations in fire-proof areas, and the transportation process of hazardous chemicals are also risk points; Hazard source refers to the root cause, state or behavior, or a combination of them, which may cause personal injury and / or health damage and / or property loss.

[0031] Among them: Source refers to the physical entity that has energy or generates or releases energy, such as lifting equipment, electrical equipment, pressure vessels, etc. Behavior refers to the decision-making behavior, management behavior and operation behavior of decision-makers, managers and practitioners. State refers to the state of objects and the state of the environment, etc.

[0032] When analyzing factors that cause casualties, affect people's health, or even cause diseases in the production process, the sources of danger can be called dangerous and harmful factors, which are divided into four categories: "human factors", "material factors", "environmental factors" and "management factors". Human factors refer to dangerous and harmful factors from the personnel themselves or of a man-made nature in production activities; material factors refer to dangerous and harmful factors in machinery, equipment, facilities, materials, etc.; environmental factors refer to dangerous and harmful factors in the production environment; and management factors refer to dangerous and harmful factors caused by lack of management and management responsibility.

[0033] In another embodiment provided by the present invention, in step S2, the risk model consists of a vision module and a pre-judgment module. The vision module is established based on vision detection technology. The vision module is divided into an image feature extraction unit and a text link unit. The image feature extraction unit is used to identify and extract the collected data. The image feature extraction unit extracts and identifies image information through five methods: noise reduction processing, image enhancement, color image conversion to grayscale image, grayscale image conversion to binary image, edge detection / splitting, and histogram matching / contour matching. The text link unit selects appropriate text information to match the extracted image according to the results extracted by the image feature extraction unit. Risk data is stored in the text link unit. The pre-judgment module is established based on machine learning technology. The pre-judgment module pre-judges risks according to the following calculation formula. Among them, the formula is as follows:

[0034] Among them, S is the risk probability. When S≥1, the risk is high and key attention and control are required. When 0<S<1, the risk is relatively low and attention is needed. When S = 0, there is no risk. N is the number of hazard sources and risk points in a certain area within the factory, M is the maximum allowable number of hazard sources and risk points in a certain area within the factory, n is the level of accident hazard degree when hazard sources and risk points in a certain area within the factory occur together, and m is the maximum allowable hazard degree when hazard sources and risk points in a certain area within the factory occur together. There are three risk levels in step S2, which are specifically as follows: Level 3 risk: No risk; Level 2 risk: Slightly dangerous, attention is needed; Level 1 risk: Control and rectification are required.

[0035] In another embodiment provided by the present invention, in step S3, the monitoring device selects an AI intelligent monitoring camera. In step S3, the layout of the monitoring device should fully cover the key areas of the factory. In step S3, the network connection device selects an industrial control computer.

[0036] In another embodiment provided by the present invention, in step S5, there are two ways to send alarm information. One is to directly send problem information through the factory's speakers and alarms, and the other is to contact the factory personnel by means of text messages, phone calls, etc. The supervision and handling in step S5 include four methods: risk avoidance, loss control, risk transfer, and risk retention.

[0037] It should be noted that: Risk avoidance: Consciously abandon risk behaviors and completely avoid specific loss risks. This method is applicable to situations where there is extreme aversion to risks or when there are other target solutions with lower risks.

[0038] ‌‌Loss Control‌: Make plans and take measures to reduce the possibility of loss or reduce actual loss. Pre-event control is mainly to reduce the probability of loss, while in-event and post-event control is mainly to reduce the actual loss.

[0039] ‌Risk transfer‌: Risk is transferred to other participants or insurance companies through contracts. The main forms include contract transfer and insurance transfer.

[0040] ‌‌Risk retention‌: that is, risk assumption. When facing losses, the investment entity uses the funds it can gather to make up for the losses. Planned self-insurance and unplanned self-retention are the two main forms.

[0041] Working principle: Collect factory data, then process and analyze the data to obtain risk data; according to the risk data, establish a risk model through visual inspection technology and machine learning technology. The visual module of this process can identify and extract the collected data, match the text information with the extracted image, and establish the risk level according to the situation in the factory; select suitable monitoring equipment and fix it in a suitable position in the factory. The monitoring equipment uses AI intelligent monitoring cameras. The layout of the monitoring equipment in step S3 should fully cover the key areas of the factory, and set up a network connection device. The monitoring equipment is connected to the main control server through the network connection device; the monitoring equipment implements the collection of data information, the risk model identifies and extracts the data, and estimates the risk based on the identification and extraction results; when the model estimates that there is a risk, an alarm message is issued to remind workers to control and supervise the estimated risks in the factory.

[0042] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A smart factory risk management method based on a visual large model, characterized in that: It includes the following steps: S1. Collect data and process and analyze the data: Collect factory data, and then process and analyze the data to obtain risk data; S2. Build a risk model and set up risk levels: According to the risk data, establish a risk model through visual detection technology and machine learning technology, and set up risk levels according to the situation in the factory; S3. Arrange monitoring devices and establish network connections: Select appropriate monitoring devices, fix them in appropriate positions in the factory, and set up network connection devices. The monitoring devices are connected to the main control server through the network connection devices; S4. Collect data information and the model predicts risks: The monitoring devices continuously collect data information, the risk model identifies and extracts the data, and estimates risks based on the identification and extraction results; S5. Send out risk information and control the risks: When the model predicts risks, send out alarm information to remind workers to control and supervise the predicted risks in the factory.

2. According to the method for risk management of a smart factory based on a visual large model according to claim 1, it is characterized in that: The risk data in step S1 includes hazard sources and risk points in the factory.

3. The method for risk management of a smart factory based on a visual large model according to claim 1 is characterized in that: The risk model in step S2 consists of a visual module and a prediction module. The visual module is established based on visual detection technology. The visual module is divided into an image feature extraction unit and a text link unit. The image feature extraction unit is used to identify and extract the collected data. The image feature extraction unit extracts and identifies image information through five methods: noise reduction processing, image enhancement, conversion of color images to grayscale images, conversion of grayscale images to binary images, edge detection / splitting, and histogram matching / contour matching.

4. The method for risk management of a smart factory based on a visual large model according to claim 3 is characterized in that: The text link unit selects appropriate text information to match the extracted image according to the results extracted by the image feature extraction unit, and the risk data is stored in the text link unit.

5. The method for risk management of a smart factory based on a visual large model according to claim 3 is characterized in that: The prediction module is established based on machine learning technology. The prediction module predicts risks according to the following calculation formula. The formula is as follows: Where S is the risk probability. When S≥1, the risk is high and key attention and control are required. When 0<S<1, the risk is relatively low and attention is required. When S = 0, there is no risk. N is the number of hazard sources and risk points in a certain area in the factory, M is the maximum number of hazard sources and risk points allowed in a certain area in the factory, n is the level of the degree of accident harm caused by hazard sources and risk points occurring together in a certain area in the factory, and m is the maximum allowed degree of accident harm caused by hazard sources and risk points occurring together in a certain area in the factory.

6. The method for risk management of a smart factory based on a visual large model according to claim 1 is characterized in that: There are three risk levels in step S2, which are specifically as follows: (1) Level 3 risk: No risk; (2) Level 2 risk: Slightly dangerous, attention is required; (3) Level 1 risk: Control and rectification are required.

7. The method for risk management of a smart factory based on a visual large model according to claim 1 is characterized in that: In step S3, the monitoring device selects an AI intelligent monitoring camera, and the layout of the monitoring device in step S3 should fully cover the key areas of the factory.

8. The method for risk management of a smart factory based on a visual large model according to claim 1 is characterized in that: In step S3, the network connection device selects an industrial computer.

9. The method for risk management of a smart factory based on a visual large model according to claim 1, characterized in that: There are two ways to send out alarm information in step S5. One is to directly send out problem information through the speakers and alarms in the factory, and the other is to contact the factory personnel by means of text messages, phone calls, etc.

10. The method for risk management of a smart factory based on a visual large model according to claim 3 is characterized in that: The regulatory processing in step S5 includes four methods: risk avoidance, loss control, risk transfer, and risk retention.

Images