Combustion process multivariable control method for CFBB (circulating fluidized bed boiler)

Abstract

The invention discloses a combustion process multivariable control method for a CFBB (circulating fluidized bed boiler), which is realized in the following procedures: in each control period, collecting operational parameters of the boiler through data collecting equipment and storing the operational parameters in a data storage module; utilizing the history data in a memorizer to on-line identify the CARIMA model and present P step future moment predominant values such as process output variable main steam pressure, material bed temperature and flue gas oxygen content through a model on-lineparameter identification module of GPC (generalized prediction control); performing error compensation to the process future moment prediction output through an error estimation module of an LSSVM (least square support vector machine); and referring the reference trace obtained by a trace generator, performing rolling optimization in GPC for the process future moment prediction output, and calculating through the optimized algorithm to enable the process actual output to reach the set value. The method provided by the invention solves the time varying problem of the model parameter, and enables the control system to have stronger robustness.

Description

technical field [0001] The invention relates to a control method for the combustion process of a circulating fluidized bed boiler, specifically a multivariable generalized predictive control method based on least squares support vector machine error compensation, involving automation technology and pattern recognition technology, and used for the boiler combustion process control. Background technique [0002] The circulating fluidized bed boiler (CFBB) is developed on the basis of the bubbling fluidized bed boiler. The characteristics of desulfurization in the furnace and low-temperature combustion inherent in the fluidized bed combustion method greatly reduce NO X Emissions make the fluidized bed boiler a widely used power equipment for steam production. While the capacity of circulating fluidized bed boilers is constantly increasing, there are problems of low automation level of combustion system and low quality of automatic control. Circulating fluidized bed boilers i...

AI Technical Summary

Problems solved by technology

[0005] Model predictive control is another widely used control strategy in the field of industrial processes. There have been many literatures on the application of predictive control in the boiler combustion process, but most of them focus on single process variables such as superheated steam temperature, reheated steam temperature The univariate control does not solve the coupling problem of the multivariable process well, nor does it give full play to the characteristics of the predictive control to directly deal with the multivariable process.

Images