Safety production is the eternal theme of social development and the true meaning of all work.
For industrial production enterprises, due to strong business continuity, complex systems, a large number of production equipment are closely connected and coupled, and have the characteristics of high power and high operation speed. Safety production is a necessary condition to ensure the personal safety and health of employees, equipment and facilities from damage, the environment from damage, and the smooth progress of production and operation activities.
As the petroleum and petrochemical industry is a high-risk industry with many dangerous and hazardous factors, the safety production situation is still very serious.
The role of artificial intelligence in the field of safety production has been significantly improved
By means of artificial intelligence technology, the company accurately grasps the industry needs of “high-precision quality detection and wide-ranging safety management”, and applies artificial intelligence technologies such as machine vision, posture recognition, abnormal behavior analysis and early warning to realize real-time monitoring, automatic problem detection and active early warning in terms of safety prevention, supervision and implementation, quality detection and production process management, thus improving the dilemma of relying on the naked eye or “remote water can not save near fire” in the past, To ensure safe and efficient production, appropriate labor distribution and low cost, in order to assist industrial enterprises in “cost reduction, efficiency increase, safe production”, and other intelligent applications, it has gradually played an important role in the field of industrial production safety, changed the previous “post-processing” mode of safety management, and shifted to the scientific management mode of pre identification, analysis and control of hazards, and finally realized pre-control and prevention first, The goal of moving the pass forward is to nip in the bud.
Generally, petrochemical enterprises have built video monitoring systems by stages and in batches over the years, basically realizing that each production device and key parts have been installed with monitoring cameras. Traditional machine vision inspection (such as comparison method) solves some problems of manual visual inspection, such as the inability to do it well and the high cost of doing it. However, there are still potential safety hazards:
1. In the process of on-site operation supervision, due to various reasons such as human and material resources, the relevant management personnel in each link sometimes cannot come to the construction site to supervise and monitor, review and confirm the work permit, or stay on the site for a symbolic short time, which is difficult to meet the requirements of the current operation permit system and specifications.
2. Although the whole process management of electronic approval of work permits has been realized, it is difficult to supervise the non-standard behaviors of workers on the construction site and the illegal use of equipment and facilities. Even if there is video monitoring, it is generally limited to manual identification and monitoring in the control room, and even only the main production devices and key parts can be monitored, It cannot meet the requirements of monitoring in case of non-standard operation in any production area.
3. It is also very important to supervise the Contractor’s labor personnel. Before the Contractor’s construction personnel enter the site, safety education and training, vocational skill review and other work have been carried out. However, there are also potential safety hazards such as replacement when they actually arrive at the site for construction.
Compared with traditional machine vision detection methods, the detection method based on artificial intelligence will reduce the dependence on external factors such as illumination, placement position, transmission rate, etc., especially the neural network learning for a large number of images of some behaviors that are difficult to recognize. With sufficient training, it will provide a higher recognition accuracy for the main and non main features of various behaviors and objects.
How to implement artificial intelligence in the field of safety production
Through the establishment of a set of intelligent monitoring and management system for safe operation, the monitoring and management of the operation site will be strengthened. At the same time, the management method of territorial management will be implemented to discover the potential safety hazards in each production link in time, so as to analyze and deal with the potential hazards in time and finally achieve safe production.
The specific application scenarios are as follows:
1. The intelligent monitoring system for safe operation based on artificial intelligence is combined with the work permit management system to realize the real-time monitoring of the whole process of on-site operation, and the monitoring and management of intelligent identification and identity verification of the supervisor, ticket reviewer, operation applicant and the construction personnel dispatched by the contractor at the construction site. At the same time, it can also monitor whether there are dangerous behaviors of people in the operation area Whether there is any non-standard behavior such as unauthorized person or object crossing the boundary.
2. The machine vision recognition technology based on artificial intelligence is used to replace the traditional artificial vision recognition method to realize the automatic intelligent recognition and early warning function.
3. It can view the video images of each monitoring point in real time through the client management software system, and call, warn and broadcast the remote video of key monitoring points.
4. Through the system, you can view various recent underground operations, hidden dangers and video alarm processing through video playback.
5. Realize the hierarchical network architecture, and the superior monitoring center can view and manage all video images within its jurisdiction.
How to design an AI based intelligent safety production monitoring system
The safety operation intelligent monitoring system based on artificial intelligence undertakes the important tasks of safety command and control, communication, data collection, uploading and sharing in petrochemical enterprises. It is the key and link of enterprise safety production and management informatization. Its design principle must ensure that the whole system has the characteristics of high reliability, strong stability, advanced technology, friendly man-machine interface, simple operation, convenient maintenance and convenient upgrading.
1. Meet the needs of group monitoring management application, fully consider the short-term, medium-term and long-term development of the company in the future, put forward specific construction schedule, and provide efficient and high-quality system wide technical support services.
2. The existing investment shall be protected, and the existing monitoring equipment installed and used shall be treated in a compatible manner in the construction of the new system as far as possible, so as to save construction resources and lengthen equipment operation cycle.
3. The petrochemical plant area is a high-risk area, so the active equipment shall be explosion-proof equipment complying with national standards, which shall not only improve the safety level in the plant area, but also meet the requirements of video image acquisition.
4. The system is easy to operate, and can easily control various equipment of the system. The operation is visual and simple, and there is no need to memorize various cumbersome functions. In addition, the control panel, media player and video playback query are integrated in the same client software interface, and the monitoring system terminal can realize all operations through the same single software, which is convenient for the client operation.
5. The interface is open and can be seamlessly connected with other related systems.
The safety production monitoring and management system based on artificial intelligence is different from the traditional security system. It is not only video monitoring, but also takes the deep neural network computer vision AI technology as the core, and uses machine vision to replace the human visual supervision, so as to truly liberate human resources and achieve 24-hour seamless supervision, greatly save human resources, and make the disposal methods more efficient and diversified.
The traditional machine learning technology often uses the original form to process natural data, and the learning ability of the model is greatly limited. To form a pattern recognition or machine learning system, it often requires considerable professional knowledge to extract features from the original data (such as the pixel value of the image) and convert them into an appropriate internal representation.
Deep learning has the ability to automatically extract features, which is a kind of representation oriented learning. Deep learning allows multiple processing layers to form complex computing models, so as to automatically obtain the representation of data and multiple levels of abstraction. These methods have greatly promoted the development of speech recognition, visual object recognition, object detection, drug discovery and genomics. By using BP algorithm, deep learning has the ability to discover the hidden complex structure in large data sets.
Evolution of safety production management from administrative management to data driven
Facts have proved that emerging technologies such as the Internet of things, big data and artificial intelligence are undoubtedly a powerful starting point for enterprises to achieve all-round safety production management. Compared with “system management”, they can obtain measurable benefits through predictable investment. Of course, these emerging technologies have also brought confusion to the traditional safety production managers, that is, how to apply them in the safety production management system. Jiyun technology puts forward “Data-Driven safety production management”. Through hierarchical thinking, based on the unified industrial Internet platform, it implements safety production management solutions from the three levels of equipment, production and operation.
Equipment reliability management based on data drive. At the equipment level, the scheme builds a digital twin model of equipment, models the data of assets, organizations, processes and processes, forms management objects at different levels, and realizes the configuration of attributes such as indicators, dependencies and data sources, so as to effectively manage the basic data of equipment connected to the platform. On the basis of these basic data, the scheme develops predictive maintenance applications for key production equipment of the enterprise to reduce the failure rate and improve the reliability of the equipment.
Key process flow management based on data driving. At the production level, the scheme collects various data in the production process, and improves the yield of finished products through refined multi-dimensional, long-term, process based and artificial intelligence data analysis. For various key indicators that need to be monitored in production, the scheme calculates the corresponding monitoring indicators through the model according to the characteristics of the indicators, sets the monitoring strategy, and alerts the exceptions that violate the strategy.
Based on data-driven safety risk management and control decision and emergency command. The scheme establishes a full effect emergency command and safety production mechanism, comprehensively perceives the security risk situation based on big data, and forecasts the possible major emergencies by combining the statistical analysis of accident frequency, unit and other characteristics; Unified configuration perspective, break through the information barrier of control system, and realize real-time production and operation monitoring; The system also realizes business management, hierarchical early warning and alarm functions, environmental protection management, etc.
Data is the blood of equipment, production and operation. The scheme builds a converged “data center” by collecting equipment data, production data, external data, etc. Based on this integrated “data center”, the unified monitoring of equipment operation status can be realized, and the predictive maintenance of key equipment and production process optimization can be realized by combining big data and artificial intelligence. On this basis, the whole Bureau Management at the overall operation level can be realized, and finally the purpose of improving the safety production management level can be achieved.
Responsible editor: CT