Optimization Study of Heater Water Level Finding in Thermal Power Units Based on Fuzzy Weighting and LSTM Deep Learning

In order to solve the problems of low efficiency, over-reliance on experience and insufficient adaptability under unknown working conditions, which generally exist in the heater water level control when thermal power units are flexibly peaking, combining the big data processing technology and deep learning, we firstly carry out the correlation analysis of each parameter that affects the heat return system, and according to the Pearson’s correlation coefficient, we identify the key parameters that have a significant impact on the energy consumption of the heat return system. For these important parameters, the operating state of the heater system is divided by Fuzzy C-means (FCM) clustering algorithm, and the Long Short-Term Memory (LSTM) neural network is used to construct the optimal value model of the heater water level for the divided typical conditions, and based on the fuzzy weighted real-time condition matching Based on the fuzzy-weighted real-time working condition matching mechanism, the real-time data of the heater system under different boundary conditions are matched with the model to complete the optimization of the heater water level.