Two significant motivations for continuing development in early process detection problems are technological plant safety and dependability. To avoid economic losses in oil exploration, faults in Field Digitizing Units (FDUs) instruments must be detected. The creation of algorithms that can detect process issues before they reach their threshold is a big task, and trend checks associated with a particular measured value are typical. The efficiency of the Deep Neural Network (DNN) technique employing Matlab and a lowcomputational power device, such as the Raspberry Pi 4, for drift fault detection in FDUs is evaluated in this article. The DNN classifier is among the deep learning algorithms being studied. The FDUs instruments provided the data for this experiment. In training and testing data, the six features (Distortion, Noise, Common-Mode Rejection (CMRR), Gain Error, Phase Error, and Crosstalk) were extracted from free fault and faulty FDUs. The trained model has been offline tested, with the model being used to detect drift faults using FDU performance. Accuracy, specification, precision, recall, and F-measure were used to determine the efficiency of the classifier, with 99.7% accuracy in the DNN with Matlab and 98% accuracy in the DNN with Python.