Published on: 2026-09-23
Source: Peter the Great St. Petersburg Polytechnic University –
An important disclaimer is at the bottom of this article.
Scientists at the Peter the Great St. Petersburg Polytechnic University (SPbPU) have developed a hybrid architecture combining a digital twin of a combined heat and power plant (CHP) and artificial intelligence methods. It selects the optimal operating modes of the plant in about 15 seconds instead of several minutes — 16 times faster than calculation using a digital twin and traditional mathematical methods. This speed makes it possible to optimize the modes of several power plants simultaneously from a single dispatch center.
The goal of the scientific team’s work is to create a system for optimizing the life cycle of power plant equipment using predictive analytics tools. By life cycle optimization, scientists mean a multi-criteria approach aimed simultaneously at extending the equipment’s service life, increasing the efficiency of the CHP’s production and sales activities, and improving its environmental performance. In Russia, there are already software solutions that cover individual aspects of thermal power plant operation—for example, only optimizing operations on the wholesale electricity and capacity market or only assessing the technical condition of equipment. However, a comprehensive task of optimizing the entire life cycle has not yet been implemented in any software suite. Large foreign corporations are approaching such ambitious tasks, but as practice shows, their solutions do not work in the Russian market due to the absence of many calculated parameters and the complexity of the thermal schemes and modes of domestic plants.
This year, the team achieved several significant results. The first is a module for automatic intelligent reparameterization of the digital model, which is the core of the CHP digital twin: the computational models are now calibrated automatically and take into account degradation or other changes in the equipment’s condition, relying on indirect parameters that characterize its technical state. The second is the implementation of hybrid algorithms for optimizing the power plant’s operating modes, combining the digital twin and artificial intelligence.
The project leader, Associate Professor of the Higher School of Nuclear and Thermal Power Engineering at SPbPU Irina Anikina, cited as an example a case of optimizing the operational modes of a CHP plant with five power units equipped with T-100 turbines: The digital twin of this plant generates as a dataset for surrogate models all possible operating modes and technical condition variants of the power equipment—in our example, about 280,000 states. Then, information about the current mode and condition of the plant is fed into the hybrid architecture, and the optimizer finds a solution in fractions of a second. To avoid errors related to modeling inaccuracies, the solution proposed by the optimizer is sent back to the digital twin for result verification, and additionally, control modes are simulated and the correctness of the proposed solution is checked. This method significantly accelerated optimization: instead of several minutes when using the digital twin and classical mathematical methods, it now takes about 15 seconds. That is, the calculation speed increased 16-fold.
For the station staff, the result of this work looks like a hint system that helps simultaneously meet several requirements: the equipment must be reliable and require repairs as infrequently as possible, fuel consumption should be minimized while maximizing revenue from selling thermal and electrical energy, and environmental harm must be minimal. Scientists compare such a task to the long-term operation of a freight truck: it is important that the vehicle does not break down and is handled gently, but at the same time consumes the minimum amount of fuel and maintains acceptable parameters in terms of speed, harmful emissions, and driver comfort. Moreover, it is necessary to consider not only the characteristics of the vehicle itself but also external factors: traffic jams, traffic light operation modes, route terrain, weather conditions, and fuel composition.
According to researchers’ estimates, for the CHPP in the given example, operating under the modes proposed by the system would have increased marginal income by one third over six months.
The system for comprehensive optimization of the life cycle of power plant equipment is extremely necessary for all operational stations in our country and other countries with developed district heating. It is time for us to move away from regulatory parameters in repairs and efficiency assessment of operating modes and accept the fact: equipment repairs should be made not according to schedules, but based on condition, and the optimal condition should be ensured by the modes from our suggestion system. Now is the time to treat equipment, resources, and the environment carefully. Using such a system will address issues of import substitution, the high cost of repair and restoration of equipment, uneconomical and environmentally unfriendly operating modes,
The development is based on digital twins containing classical mathematical models of CHP plants, which reconcile balances for individual equipment and for the plant as a whole, as well as methods of machine learning and artificial intelligence. The obtained results can be scaled to combined heat and power plants in Russia and neighboring countries, but for the digital twin to function effectively, it is necessary to ensure data transmission from the automated process control system (APCS) beyond the plant via special industrial protocols or to organize daily manual uploading of CHP operating parameters.
Individual modules of the system — forecasting the technical condition of equipment, improving the operating modes of the plant, and optimizing environmental parameters — have already been tested in real conditions. The team’s next task is to move towards multi-criteria efficiency improvement that takes all these modules into account, and to create a high-quality suggestion system for the CHP personnel: regarding current and forecasted modes, equipment shutdown for repair, and the execution of other regulatory activities.
The scientific work is carried out with the support of the Development Program of Peter the Great St. Petersburg Polytechnic University for 2025–2036 under the implementation of the “Priority-2030” program (national project “Youth and Children”).
Please note; This information is raw content obtained directly from the information source. It is an accurate report of what the source claims and does not necessarily reflect the position of MIL-OSI or its clients.