The battery undervoltage fault is common in the field operation of smart meters. Due to the diverse causes of battery undervoltage and reliability data sources, it is difficult to carry out quantitative evaluation of battery undervoltage in engineering, and there is still a lack of engineering-applicable clock battery undervoltage evaluation and prediction methods. This method sorts out the reliability logic relationship of each component unit from the clock battery circuit level and constructs a system reliability model of the clock battery circuit for the battery undervoltage fault mod, then a time-varying weight model for system life distribution and a method for undervoltage prediction of clock battery based on system reliability synthesis theory are innovatively proposed. It realizes the comprehensive utilization of unit performance degradation data, life data and life distribution information to predict the trigger time of clock battery undervoltage event and solves the difficult problem of quantitative evaluation of the battery undervoltage of clock of electricity meter under the condition of diversity of reliability information of the constituent units. Finally, an engineering example is used to verify the effectiveness of this method.