Temperature Prediction of Nuclear Chemical Exhaust Gas Based on Multi-Kernel Relevance Vector Machine

The temperature of the exhaust gas from a nuclear chemical plant is directly related to the compliance of the filtered gas during the treatment process, however, due to the strong radioactivity of the exhaust gas, it is unable to measure the temperature of the nuclear chemical exhaust gas inside the pipeline directly with the instrumentation, In this paper, a method based on the Relevance vector machine (RVM) with multiple kernel functions is proposed to predict the exhaust gas temperature inside the pipeline by measuring the temperature of the outer wall of the pipeline. The proposed method has a sparser solution and better model generalization ability, and the prediction model of the temperature of the outer wall of the pipeline and the temperature of the exhaust gas inside the pipeline is established by using the actual data. The experimental results show that the method can predict the temperature of the exhaust gas inside the pipeline with higher accuracy and better fitting, which is of practical significance for the temperature monitoring during radioactive waste gas treatment.