Gas-steam combined cycle performance prediction based on neural network model with segmented approach
摘要
Huaneng Beijing Power Plant introduced a plate heat exchanger (PHE) into the HRSG of a GSCC system to boost heat recovery. However, the modification’s impact on the GSCC’s overall performance was uncertain. To evaluate how the utilization of the PHE has affected the system, a CNN-SA-GRU neural network model deployed with segmented approach was used to simulate the GSCC system. The models predicted pre-modification performance of the system under operational conditions of post-modification period, and the results were analysed and compared with the actual output of the post-modification system. The results showed a slight gas turbine output decrease but a boost in steam turbine output, while the recovered heat from PHE itself played the dominating role lifting the GSCC system’s combined cycle efficiency. In general, the introduction of the PHE increased the average combined cycle efficiency from 81.18 % to 86.61 %, and brought a reduction of 2.82 MJ/(kW·h) in heat rate.