Sensitivity Analysis of Gas Consumption in Gas Turbine Combined Cycle

In order to analyze the specific impact of key parameters of an internal combustion engine on its economy, a gas consumption sensitivity analysis model based on a deep feed-forward neural network is proposed. It is capable of evaluating the degree of influence of each parameter on gas consumption. In this paper, a hybrid optimization algorithm combining ant colony optimization and particle swarm optimization is used to effectively train and optimize the deep feed-forward neural network. A global sensitivity analysis is performed using the model to screen out several parameters that have a large impact on gas consumption. Local sensitivity analyses of these several parameters are performed separately, revealing the quantitative relationship between them and gas consumption. Based on the analysis results, useful suggestions are made for the operation optimization of the gas turbine combined cycle unit.