Research on Modeling Method of Variational Modal Decomposition of Boiler Combustion System
Establishing an accurate combustion system model is the basis for achieving boiler combustion optimization.Aiming at the problem of insufficient multi-step prediction accuracy of existing modeling methods,a modeling method of variational modal decomposition(VMD)of boiler combustion system is proposed.The method converts the traditional single model prediction method into a hybrid model prediction method based on data decomposition.Firstly,the VMD is used to decompose the output signal into multiple regularly varying smooth sequences to reduce the complexity of the data.Then,for each modal signal,the prediction model is built by using online least squares support vector machine algorithm and time series prediction algorithm respectively.Finally,the final prediction results are obtained by synthesis.Simulation results show that the method effectively improves the single-step and multi-step prediction accuracy of the combustion model,which can lay a good foundation for realizing the closed-loop combustion optimization control of the boiler.The method helps to solve the problem of conventional combustion system modeling methods that only focus on single-step prediction errors but neglect multi-step prediction errors.
万方数据
维普
