This paper tackles the cybersecurity challenges in smart energy systems using the IEC 61850 standard. We propose a real-time anomaly detection algorithm for identifying Denial of Service (DoS) attacks in networks with Intelligent Electronic Devices (IEDs). The algorithm, based on an Autoencoder neural network, analyzes network traffic to detect anomalies via reconstruction errors. Tested on a dataset with both normal and DoS attack traffic, the algorithm achieved optimal results, with the 90th percentile showing the highest F1-Score and perfect Recall, ensuring no anomalies are missed. The findings highlight its effectiveness in enhancing the cybersecurity of smart energy networks.