Русская версия English version

Improving the accuracy of turbine plant efficiency assessment by neural networks

V.A. Gorbunov, N.A. Lonshakov, O.Y. Nagornaya, A.A. Belyakov

Vestnik IGEU, 2017 issue 4, pp. 5—12

Download PDF

Abstract in English: 

Background: Improving the efficiency of thermal mechanical equipment is now part of the global trend towards optimization and lowering operating costs. At present, the efficiency of thermal mechanical equipment is assessed by several indicators such as specific power consumption, performance, and power. However, the error of the engineering methods used to calculate these parameters amounts to 20 %, with the required maximum error for solving optimization problems and predicting operating conditions being 5 %. That is why the goal of this study is to develop techniques of improving the accuracy of assessing performance indicators of thermal mechanical equipment.

Materials and methods: We propose to use the apparatus of neural networks trained on standard instrumentation archives data. The choice of neural networks is due to the fact that, in comparison with other statistical methods, they do not require knowledge of the dependences of performance indicators on input parameters. To demonstrate the capabilities of this method, we use operational data taken during the operation of the top-pressure recovery turbine (TPRT) and the turbine feed water pump (TFP).

Results: We have constructed neural network models for efficiency indicator assessment and by analyzing them shown that for TPRT the greatest influence on the power, specific energy consumption and efficiency of the gross blast furnace is exerted by the flow of blast furnace gas. For example, with a change in the flow rate from 678,000 to 988,000 m3/h, the increase in electric power is from 0,02 to 0,75 MW. The greatest effect on the developed power is made by pressure and steam flow to the drive turbine, a change in which within the investigated range can lead to an increase in power by 2,02 MW.

Conclusions: The reliability of the results received is confirmed by the verification of neural networks based on data not participating in the learning process. The error in assessing the performance indicators does not exceed 3 %. The obtained results are used in the process of operation for analysing the efficiency of plants, for the compilation of parameter tables, for obtaining technically justified norms of energy consumption, for improving the accuracy of turbine efficiency parameter calculation.

References in English: 

1.    Ipatov, P.L. Regional'nye aspekty otsenki ekonomicheskoy effektivnosti AES [Regional aspects of nuclear power plant economic efficiency assessment]. Izvestiya Rossiyskoy akademii nauk. Energetika, 2008, no. 4, pp. 3–11.

2.    Ismagilov, T.S. Metody resheniya zadachi prognozirovaniya v energetike [Methods for solving the forecasting problem in the energy sector]. Vestnik UGATU, 2010, no. 4, pp. 93–96.

3.     Gorbunov, V.A. Ispol'zovanie neyrosetevykh tekhnologiy dlya povysheniya energeticheskoy effektivnosti teplotekhnologicheskikh ustanovok [Use of neural network technologies to increase power efficiency of thermal power units]. Ivanovo, 2011. 475 p. (in Russian).

4.    Kartashev, A.L., Martynov, A.A. Matematicheskoe modelirovanie i optimizatsiya struktury techeniya v stupeni radial'no-osevoy turbiny mikrogazoturbinnoy ustanovki  [Mathematical Modeling and Optimization of Flow Structure in Francis Turbine Stage of Microturbine Power Plant]. Vestnik YuUrGUSer. Mashinostroenie, 2015, vol. 15, no. 3, pp. 28–36.

5.     Santoso, N.I., Tan, O.T.  Neural net based real-time control of capacitors installed on distribution systems. IEEE Trans. Power. Deliv, 1990, no. 1, pp. 266–272.

6.    Nagornaya, O.Yu., Gorbunov, V.A. Ispol'zovanie neyrosetevogo podkhoda dlya polucheniya rezhimnykh kart raboty turbiny GUBT-25 [The use of the neural network approach for obtaining parameter tables of the GUBT-25 turbine]. Vestnik IGEU, 2006, issue 4, pp. 64–66.

7.     Ponomarev, V.S., Finayev, V.I. Primenenie adaptivnykh regulyatorov na osnove neyronnykh setey v energetike [Application of adaptive regulators based on neural networks in power engineering]. Izvestiya YuFU. Tekhnicheskie nauk, 2008, no. 7, pp. 164–169.

8. Dunaev, V.A., Lonshakov, N.A., Gorbunov, V.A. K voprosu o povyshenii effektivnosti i bezopasnosti ekspluatatsii teplomekhanicheskogo oborudovaniya AES [On the issue of improving the efficiency and safety of operation of thermal mechanical equipment of NPPs]. Global'naya yadernaya bezopasnost', 2015, no. 2(15), pp. 63–70.

Ключевые слова на русском языке: 
турбина, турбоустановка, показатели энергоэффективности, газовая турбина, коэффициент использования установленной мощности, нейросетевые технологии, показатели эффективности турбин, повышение эффективности турбоустановок, увеличение выработки электроэнергии, турбопитательный насос
Ключевые слова на английском языке: 
turbine, turbine plant, energy efficiency indicators, gas turbine, installed capacity utilization factor, neural network technologies, turbine efficiency indicators, turbine efficiency increase, increased power generation, turbine feed pump
The DOI index: 
Downloads count: