<p>In a complex and changeable low-voltage distribution network environment, conventional positioning algorithms may suffer from insufficient accuracy or even misjudgment due to factors such as a complex network structure, numerous line branches, and significant load variations. To address this, an automatic location algorithm is proposed for identifying key nodes of line loss in low-voltage distribution networks. The missing value range is estimated using the nearest neighbor algorithm, and the interpolation value is dynamically adjusted to fit the data range. Considering the changes in the operating interval of the distribution network after integrating distributed power, a reward-and-punishment mechanism is introduced to adjust the allocation of line loss. A recurrent neural network (RNN) is employed to extract the characteristic patterns of line loss, and the analytic hierarchy process is applied to optimize the selection of line loss indicators. Furthermore, a correlation measurement method is utilized to compute real-time anomaly location timing, and a dynamic measurement space is established. Experimental results demonstrate that the proposed method achieves a positioning error of less than 0.2, with the predicted transformer iron loss (207.1 kW) being the closest to the actual value (207.0 kW). The normalized value is approximately 0.2, and the line loss increase rate matches the real value at around 10.24%.</p>