Space missions are critical systems that must cope with extreme conditions such as temperature changes, radiation and vibration. Due to the complexity of their structure and operation, these systems are designed in such a way that they can mitigate errors and handle critical situations. The only available communication link and way by which the ground station can monitor the health of the satellite and act upon possible failures is status telemetry. This paper presents the implementation of a machine learning-based anomaly detection system for satellite telemetry; the data used corresponds to a NASA public database of telemetry channels from the Soil Moisture Active Passive (SMAP) mission. In order to get all the possible information from the signals, a feature extraction phase was carried out defining three detectors based on recurrent neural networks (RNN), moving average (MA) and Fourier transform, respectively. An AdaBoost algorithm was trained to perform anomaly classification. The proposed system was evaluated using data partitioning into training and test sets. The results show that we can achieve performance comparable to the original anomaly detection method under the same dataset using a supervised method which incorporates knowledge about the frequency, magnitude and waveform of known anomalies.