In order to solve the problem that the traditional industrial control methods
cannot control the heating flow and water temperature in a timely and
effective manner due to the high delay and complex coupling characteristics
of the urban central heating system, the authors propose deep learning-based
data processing and management for thermal heating systems. The author
analyzes the non-ideality of district heating system and its influence on
the application of deep learning technology, and gives solutions,
respectively, finally, a primary side regulation scheme of district heating
system based on deep learning and automatic control technology is proposed
as a whole. The experimental results show that, by comparing the water
supply temperature predicted by the equipment model of the primary side heat
station with its actual measured value, the mean square error of the
prediction results using the model directly is 1.30%, and the mean square
error after model correction is 0.094%. The secondary return water
temperature was controlled by adjusting the opening of the primary side
electric valve, the expected secondary return water temperature in the
scheme was compared with the actual secondary return water temperature, and
the mean square error was 0.102%. It is proved that the scheme can achieve
good control effect in the actual system, and the data result proves that
the scheme is feasible.