A regression-based method for estimating the number of power poles in areas of electricity consumption regulation

Utility poles play a fundamental role in overhead power networks, supporting the equipment for delivering electricity to end users. This study proposes a regression method based on a Multilayer Perceptron (MLP) neural network to estimate the number of poles in areas undergoing regularization. The methodology was evaluated using a dataset comprising over 3.5 million meter records and 850,000 poles obtained from public and private datasets. To structure the data, meter points were geographically grouped using the DBSCAN algorithm, enabling the automatic segmentation of regions densely populated with meters. Spatial features such as area, perimeter, and Minimum Spanning Tree (MST) metrics were then extracted and used as input variables for the models. The MLP achieved the best performance, reaching a Mean Absolute Error (MAE) of 4.95 and a standard deviation of 0.24. These results indicate that the proposed approach is effective and can be applied to future estimations of pole distribution in areas subject to regularization.