Method and system for predicting short-term temperature resistance values ​​based on AdaBoost ensemble learning

Abstract

By considering characteristic factors, the resistance value of the temperature measuring resistor is predicted for a certain period in the future, areas within the unit where abnormal temperatures may occur are inspected, abnormalities in the unit's temperature measuring resistance value are prevented, and the risk of unit failure is reduced. [Solution] The short-term temperature measurement resistance value prediction method constructs a temperature measurement resistance value prediction framework based on adaptive boosting (AdaBoost) ensemble learning, analyzes the correlation between each feature factor and the temperature measurement resistance value at the current time using the maximum information coefficient based on time feature factors, screens for feature factors that strongly correlate with the temperature measurement resistance value, trains each individual learner of the temperature measurement resistance value prediction framework by combining the unit's temperature measurement resistance value data and the feature factors that strongly correlate with the temperature measurement resistance value, adjusts the sample weight values ​​through an iterative process to obtain a final strong learner, and performs temperature measurement resistance value prediction according to the strong learner to obtain the prediction result.

Images