“During the flexible peak shaving process of thermal power units, the superheated steam temperature fluctuates greatly, which affects the safe and economic operation of the units. Accurately predicting its trend is crucial. A superheated steam temperature prediction model based on an improved whale algorithm optimized long short-term memory neural network is proposed. The principal component analysis method is used to select variables to eliminate data redundancy. Nonlinear convergence factor adjustment strategy, adaptive weight strategy, and dynamic spiral update strategy are used to improve the whale algorithm to improve the optimization accuracy and global optimization ability. A prediction model is established for a DC boiler in a power plant in Xinjiang, and the power plant data simulation is applied. The results show that the improved algorithm has higher prediction accuracy, and can predict the trend well during load increase and decrease periods. The average absolute error and root mean square error are reduced by 7.02% and 6.22% respectively. This improved algorithm effectively improves the accuracy of the prediction model. The sentence is:, This provides a foundation for optimizing the design of the superheated steam temperature control system.”