Deep Neural Networks have steadily gained importance due to great success in many application areas. The selection of an optimal combination of a data preprocessing technique, model architecture and training parameters is a labor-intensive process. In this bachelor thesis, we propose an algorithm, that automates the architecture search for the task of Classification of Time Series Data. Sequential architectures are found, evaluated and improved using evolutionary techniques. Compared to standard architectures in that field, the evolutionary evolved architecture achieves an accuracy improvement of up to 5.82 percent.