The invention discloses a big data mining-based power network low voltage cause diagnosis method. The method comprises the steps of performing data cleaning on initial data; integrating source data to be in a specified format; executing a clustering algorithm to classify low voltage data; marking low voltage causes: marking clustering results in the step 300 in a classified way, wherein different types represent low voltages of different causes; classifying all data until a final low voltage classification result is obtained. The method is suitable to process massive data and provides decision support for optimizing a low voltage investment scheme and determining a low voltage investment direction clearly; the manpower and material resources are saved; and the power network management efficiency is improved.