Abstract: The private capacity expansion of special transformers not only violates the interests of power companies, but also affects the safe operation of power grids. The current method for monitoring the capacity expansion of special transformers cannot meet the requirements of supervision and intelligence due to the need to manually select multiple groups of voltage and current on the primary and secondary sides for data fitting. In this paper, a data-driven method is used to carry out the monitoring research on the capacity expansion of special transformers. It is determined that the primary current, the secondary voltage deviation, the power factor, and the three-phase unbalanced current are used as the characteristics to train the load rate calculation model, and the error between the calculated value of the load rate and the monitored value is used as the basis for judging the capacity expansion. Further, a capacity expansion monitoring method based on the Long/Short-Term Memory (LSTM) network nested Fuzzy-C-Mean (FCM) clustering is proposed, and the real power consumption acquisition system data is used for the example analysis. The results show that based on the load rate calculation model of the LSTM-FCM clustering, the calculation error of the load rate is within 5%, and a single calculation time is about 0.05s, which achieves the purpose of monitoring the special changes and expansions.