The neuronet will help Unipro to optimize process of repair work

2020: Process optimization of repair work

On May 14, 2020 Center 2M, the Russian information operator, the videonanalitiki CenterVision showing possibilities of an intelligent system in PJSC Unipro, one of leaders in the sector of thermal generation of the electric power in the Russian Federation announced project completion. Automation of process of monitoring of actual output of personnel when carrying out repair work by calculation of number of employees and time of their stay in a zone of carrying out repair work became its main objective.

This project was implemented within development of the concept of operational efficiency and digital transformation of PJSC Unipro.

At the first stage tasks which the system of video analytics on the basis of artificial intelligence should execute were defined. When carrying out repair work at the enterprises using a neuronet it was necessary to count the number of the employees performing repair of the equipment and average time of their stay in a zone of carrying out repair work which was under observation.

After approval of tasks and requirements to providing result the specialists Center 2M within a month trained the system of industrial video analytics CenterVision using neural networks for recognition of personnel at manufacturing enterprises, to measurement of time of their stay in the studied zone. According to the results of carrying out pilot start Center 2M showed system operation with the final level of reliability of model of 87%.

Within accomplishment of tasks of increase in operational efficiency the prototype, created Center 2M, by means of surveillance cameras is capable to count automatically quantity and the time spent of personnel in a zone of carrying out repair that allows to carry out efficiency analysis of its work and to develop the relevant activities.

Thanks to technologies on the basis of artificial intelligence the Unipro company is going to increase efficiency of use of internal repair personnel and to reduce costs regarding the works performed by contract organizations.

We are grateful to Unipro company for the put trust. It is very important that such enterprises, large and crucial for our country, are not afraid to implement and develop the technologies based on machine vision. According to our forecasts, in 2020 the need for intellectual video analytics on domestic enterprises will grow more than by 20%. Decrease in operating costs and observance of the becoming tougher safety rules of production will become the main premises to it, even in the conditions of crisis. Application of video analytics based on neural networks will not demand investments into acquisition of additional technology equipment. Using technologies of deep learning, video systems can control more precisely product quality, activity of personnel, management of technological processes. The staff of industrial enterprises can reveal potentially dangerous situations thanks to predictive analytics and use a system for investigation of incidents, speaks the company executive Center 2M Evgeny Miskevich.