This paper proposes data-based fault detection methods for an electromechanical actuator (EMA) with a brushed DC motor. The jam and winding short faults are considered in the study as the most prominent EMA faults. The fault detection is based on evaluating the properties of the motor current, considering the basic electromechanical parameters of EMAs. The main advantages are a non-intrusive approach utilising a commonly accessible motor current measurement, simple configurability, and the ability to detect faults under varying operation modes of EMA, including changes of speed, load, or movement profiles. The proposed methods have been evaluated with a custom testing system, and the results have proven the performance of the proposed approach to detect faults under varying operating conditions in industrial applications.