In power plant

zirconia oxygen analyzer has the shortcomings of high maintenance cost and measurement lag.To solve this problem

this paper proposes a soft measurement method of oxygen content based on the nonparametric combination.First of all

different principles are used to establish the soft measurement models of oxygen content.From the perspective of variable correlation

kernel principal component analysis and radial base function(KPCA-RBF) model and kernel principal component analysis and support vector machine(KPCA-SVM) model are established by KPCA and SVM method.From the view of energy balance

oxygen content mechanism model is established.From the angle of time accumulation

the time series model of oxygen content is established.Finally

for comprehensive utilization of the above models' information and improving the measurement accuracy of oxygen content

taking nonparametric model as combination weight

the soft measurement model of oxygen content is established.On that basis

the test is carried out in field data of an ultra-supercritical 660 MW unit.The results show that the soft measurement model based on nonparametric combination has higher measurement accuracy and can satisfy all working conditions

compared with single soft measurement model.The predicted oxygen content can be used both as a feedforward signal to improve control quality and a dundant signal to replace the failed zirconia oxygen analyzer for uninterrupted parameter monitoring.