Study On Soft-Sensing Of Oxygen Content And Efficiency In Combustion System Of Coal-Fired Boiler

Combustion system is the core of a boiler. Its performance decidesthe economy and safety of boiler operation. Oxygen content in flue gasand boiler efficiency are two of the most important parameter related withboiler operation condition. To overcome the difficulty in themeasurement of oxygen content in flue gas and the real-time calculationof boiler efficiency, this dissertation builds soft-sensing modelrespectively. The main work is presented as follow:Firstly, a state-constrained moving window algorithm is presentedand used to build an LSSVM model of oxygen content in flue gas. Thesimulation shows the new model is better than an ordinary LSSVMmodel. And this model can be used as data backup and comparison.Secondly, with discrete boiler efficiency value calculated byanti-balance method as training data, an on-line boiler efficiency model isbuilt based on a hybrid model of PLS and LSSVM, to achieve thereal-time calculation of boiler efficiency.Thirdly, the boiler efficiency model is analyzed more deeply and a boiler optimization system is built based on PSO, with boiler efficiencyas the optimization target function. And an operation suggestion ispresented.