Research On Modeling Of Main Steam Temperature Control System Based On Historical Data

The main steam temperature object of the thermal power unit is a kind of large inertia,large delay,and nonlinear time-varying thermal engineering objects.As one of the important parameters of the safe operation control of the unit,in order to improve its control quality and ensure economic benefits,we must first understand the thermal characteristics of the main steam temperature system and establish an accurate model.Compared with the traditional modeling method,the thermal modeling method based on historical data occupies a great advantage,but most of the data modeling results can not match the parameters of the field controller,only stay in the simulation stage.Therefore,the modeling of the main steam temperature control system is divided into two parts: the actuator and the controlled object,which provides a new idea for the modeling of the main steam temperature system.In this paper,particle swarm optimization algorithm and long short-term memory algorithm are applied to the modeling of main steam temperature control system,and a main steam temperature control system model based on historical data is established for a 330 MW unit.Through the in-depth analysis of the factors affecting the main steam temperature,the principal component analysis method is introduced for the auxiliary variable selection,and the input variables of the model are optimized to ensure the validity and accuracy of the modeling results.Based on the particle swarm optimization algorithm,the mathematical model of the actuator is established,and the main steam temperature object model is divided into two parts: the leading area and the inert area.Long short-term memory algorithm modeling compares the prediction results of univariate input and multivariate input.The results show that the particle swarm algorithm is fast and can obtain an intuitive transfer function model but the model accuracy is low.The long-term and short-term memory algorithms have high precision,the algorithm is stable and can accurately reflect the characteristics of the controlled object.At the same time,the long short-term memory algorithm is large.Sample data modeling is more advantageous,and long short-term memory algorithms provide new ideas for data-based thermal modeling.