A random forest model for boiler drum water level prediction and fault diagnosis based on the characteristic variable stagger method

Abstract

In order to predict the trend of water level in the boiler drum in advance and to determine the potential fault types that cause abnormalities of water level, A Random Forest model for Boiler Drum water level prediction and fault diagnosis based on the characteristic variable stagger method is proposed. Based on the dynamic characteristics of the drum water level, the lag time difference in the actual operating conditions is identified, the characteristic variables are found upward to build a new sample set, and a Random Forest regression model is constructed to obtain the predicted drum water level, which is brought into the Random Forest fault diagnosis model to obtain the drum water level abnormal fault type. In order to verify the accuracy of the model, it was compared with Logistic Regression and Support Vector Machine models. The experiments show that the method not only can get the predicted value of Boiler Drum water level in advance, but also can quickly diagnose the potential fault types of water level abnormality, which can help the operational staff to make early prediction and take timely countermeasures for different potential fault categories to ensure the safe and stable operation of the unit.