In response to the national dual-carbon strategy

the optimized operation of heating units is of great significance to energy conservation and emission reduction. How to determine the true and reliable benchmark value of energy efficiency status is the key to operation optimization. Based on the data mining algorithm

the historical data of the unit is analyzed. Firstly

by taking load and heating flow as the characteristic quantities

the interval estimation method is applied for steady-state detection. Secondly

for the heating period and the pure condensation period of the unit

the equal interval classification method is used to divide the working conditions.Then

for the data in each steady-state operating condition

the Gaussian mixture model(GMM) algorithm model is used to select the sample with the lowest heat consumption as the reference sample. Based on the reference sample

the probability density distribution is further estimated to obtain the benchmark of the energy efficiency state under each operating condition interval. The BP algorithm is used to establish the energy efficiency state benchmark regression model

so as to establish the benchmark value working condition library under multiple working conditions of the unit. Finally

the historical data of a 320 MW heating unit is selected for method verification and online real-time monitoring. The analysis results show that

the method can optimize the operation of the unit

so as to achieve the goal of energy saving and emission reduction.