Petrochemical industry risk reduction system on the environment and animals

Petrochemical industry risk reduction system on the environment and animals

The aim of this project is to prevent hazards and reduce the negative interactions of the refinery and animals; An intelligent method based on machine learning algorithms, machine vision, deep learning and learning techniques. This method is implemented by developing technology and automating monitoring and smartening an integrated system without the need for the presence of human resources with two goals and approaches. First, reducing the risks posed by the presence of animals in refinery facilities; For example, rodents chewing wires in facilities.
Second, reduce any damage to animals, including damage from various types of pollution; As we know, the refinery always pollutes its surroundings. The emissions of these pollutants sometimes exceed the standard; At this time, humans are alerted to the danger despite alarms and leave the hazardous environment while the animals are in the environment and are harmed. It detects, then alerts a central system, where the necessary action is taken to remove the animal from the environment without human intervention.
In this method, the following data are first taught to the machine:
- Information about the identification and differentiation of different animal species.
Technical information related to the normal and abnormal limits of pollutant emissions from the refinery.
Marking areas at risk and the type of potential danger.
Using machine vision, deep learning, and machine learning algorithms, the system arrives at a network of comprehensible concepts that can detect the dangers of an animal in an environment; After detecting the animal species, it sends a message to the warning systems and the system takes the necessary action to remove the animal according to the detected species. The machine achieves this perception by collecting data received from the environment focusing on video and still images.
After the machine learning phase for data classification using learning and deep learning algorithms, a library is generated that includes a separate classification of the information we have taught the machine. Finally, the system is constantly monitoring the security of the environment and if it detects a danger, it reports and prevents its occurrence.