Modelling of thermal power unit target value based on hierarchical regression

The operation optimization of thermal power units is of great significance for energy saving and consumption reduction. Under certain boundary constraints, determining the target value of adjustable parameters is the key to optimal operation of the unit. The basic method is to analyse the historical operation data, from where select the data with the best performance indicators in all range of working condition as the target sample, to establish the model between the boundary parameters and the adjustable parameters. The most difficult step is to solve the conditional probability distribution of the performance parameters with respect to the boundary parameters. In this paper, a hierarchical regression method is proposed to slice the raw data about the performance index, and the historical data is conditionally segmented. Based on this, a modelling framework is designed for the target value of the adjustable parameters of the thermal power unit. Finally, the operation data of an ultra-supercritical 1000MW thermal power unit is selected for verification. The results show that the method has the advantages of stable, and can provide the adjustment direction for the optimal operation of thermal power units. (paper)