EPS to Introduce Artificial Intelligence in Production Processes in Hydro Power Plants, Thermal Blocks and Mines

Elektroprivreda Srbije (EPS) has opened the tender for the implementation of artificial intelligence in the production processes.

The tender worth RSD 47.7 million is split into two lots – the first pertains to the system for the prediction of the inflow at the profiles of the Drina and Lim hydro power plants, through the implementation of artificial intelligence, and the other pertains to the detection of anomalies in the operations of certain elements in the systems of production at TENT.

As said in the technical specification of the first lot, the aim of the project is to develop a system which enables the prediction of the daily inflows at the profiles of the HPPs (HPP Piva, HPP Visegrad, HPP Bajina Basta, HPP Zvornik, HPP Potpec, HPP Bistrica – Radoinja accumulation) and the flow through the hydro stations of interest (the selected stations on the Tara, Lim and Drina).

The prediction of the daily flows will be carried out on the basis of the weather forecast (precipitation and air temperature), but also based on the realization of the inflow in the past several days.

The system for predicting the inflow will consist of two main units: the part for the collection of data and the part for the prediction of the inflow. The first part includes automated services for reading the data from various sources and forming a central database, and the second part is based on a mathematical model based on machine learning for the projection of the inflow, services for the automation of the calculation and the visualization of the results of the calculation.

The system which will be developed will enable the provision of relevant information which is needed for the management of the system of accumulations and power plants in the drainage basin of the river Drina under various hydro and climate conditions (periods of medium, low and high waters), as well as technical conditions (for example, equipment overhauls), the tender says.

The system will support optimal management of the system of HPPs on the Drina and Lim in real time, so that the hydro energy production is optimized in regular hydro conditions (a better placement of electrical energy at higher prices), so that the losses of production of electrical energy during the periods of high waters are reduced and so that the damage caused by the flooding upstream and downstream from the accumulations (the Drina and the Lim) is reduced.

Also, the system will enable a more efficient work of the users, with the reduction of the consumption of work time, in planning and managing power plants and accumulations, considering that it will be fully automated.

Within the second lot of the public procurement, the System of predictive maintenance has the purpose to detect anomalies in the work of the production systems, such as turbines, generators, boilers and so on, on time before they lead to bigger problems and shutdowns, and it also has an impact on extending the life cycle of the equipment.

The specification says that it is necessary to realize the system through the implementation of a model based on AI algorithms, which would be trained with historical data provided by the information system in the production processes (TENT A, Kolubara open pit mine, PROTIS system).

After the completed training process, the model needs to be used for a quick, automatic and simple detection of anomalies.

The tender is open until November 27, and other details about the procurement (in Serbian) can be viewed HERE.

B. P.

Elektroprivreda Srbije ad Beograd

Elektroprivreda AD

EPS

Drina Lim HPP

TENT A

PK Kolubara

HPP Piva

HPP Višegrad

HPP Bajina Bašta

HPP Zvornik

HPP Potpeć

HPP Bistrica

Radoinja accumulation

PROTIS

artificial intelligence

production process management

machine learning

prediction