Stacked Auto-Encoder Modeling of an Ultra-Supercritical Boiler-Turbine System

Abstract

The ultra-supercritical (USC) coal-fired boiler-turbine unit has been widely used in modern power plants due to its high efficiency and low emissions. Since it is a typical multivariable system with large inertia, severe nonlinearity, and strong coupling, building an accurate model of the system using traditional identification methods are almost impossible. In this paper, a deep neural network framework using stacked auto-encoders (SAEs) is presented as an effective way to model the USC unit. In the training process of SAE, maximum correntropy is chosen as the loss function, since it can effectively alleviate the influence of the outliers existing in USC unit data. The SAE model is trained and validated using the real-time measurement data generated in the USC unit, and then compared with the traditional multilayer perceptron network. The results show that SAE has superiority both in forecasting the dynamic behavior as well as eliminating the influence of outliers. Therefore, it can be applicable for the simulation analysis of a 1000 MW USC unit.

Factor Analysis of the Thermal Scheme of CHP With Super Critical Steam Cycle on the Basis of Exergy Method

“…Як відомо [5][6][7], у будь-якій енергоперетворюючій системі завжди існує елемент (або сукупність елементів), зміна термодинамічних втрат в якому найбільшою мірою позначається на ефективності системи в цілому. Наприклад, у енергоблоках з суперкритичними параметрами пари -це котлоагрегати [8,9].…”

Artificial intelligence enabled efficient power generation and emissions reduction underpinning net-zero goal from the coal-based power plants

“…The model presented the merit of efficacy in modeling the complex and non-linear power production operation over the recursive least squares technique [26]. Zhang et al developed a deep neural network with a stacked autoencoder modeling approach to model power production from a 1000 MW power plant [27]. Cui et al constructed a deep neural network in addition to a deep belief network for modeling and control of a 1000 MW ultra-supercritical power generation process [28].…”

Qualitative and quantitative comparison of three modeling approaches for a supercritical once‐through generation unit

“…Znad et al 33 have presented a novel model predictive control strategy to speed up the start‐up process of a 600 MW SCPP based on a subspace state‐space identified linear model with a multi‐input single‐output structure which proved to help in more savings in fuel and water flows, and hence, fewer emissions. Furthermore, artificial neural networks (ANNs) have been extensively used to capture the SC and USC units' processes, which can be used as validated control‐oriented models 34‐36 . Various models based on NARX network have been presented to capture the dynamic processes of PCC 37 and improve its tracking capability, 38 whereas previous proposed linear models for capturing the non‐linear behavior for PCC were unable to predict it during the deviated operation from the designed operating point, 29,39 in addition, to study the non‐linear characteristics of a 600 MW SC boiler unit 40 .…”

“…Furthermore, artificial neural networks (ANNs) have been extensively used to capture the SC and USC units' processes, which can be used as validated control-oriented models. [34][35][36] Various models based on NARX network have been presented to capture the dynamic processes of PCC 37 and improve its tracking capability, 38 whereas previous proposed linear models for capturing the non-linear behavior for PCC were unable to predict it during the deviated operation from the designed operating point, 29,39 in addition, to study the non-linear characteristics of a 600 MW SC boiler unit. 40 It can be truly found that the most salient modeling philosophies for SC and USC, which are very of closely importance, are the physics-based models and identified black-box, whether this black-box be ANN or identified multi-input-multi-output (MIMO) transfer functions.…”