In this project I developed various deep learning models for anomaly detection in time series Data. The Keller Grundbau GmbH equipped some of its earth drilling machines with measuring sensors to measure parameters such as pressure, depth, aggregate power, revolutions per minute, etc. The resulting time series data could be processed by my Deep Learning models in order to detect outliers. These outliers could be detected in an unsupervised way. Two main methods were used. Firstly, we carried out a time series prediction and calculated the deviation between the prediction and the actual measured values. Secondly, we processed the time series data with a variational autoencoder. This allowed us to calculate a probability distribution in latent space. This allowed us to calculate for each individual point in the data how likely the respective value is to occur.