Sensors are vital components of the manufacturing industry due to their various applications in the production, chemical and healthcare industries. Therefore, to obtain better results and facilitate monitoring, it is of utmost importance to spot anomalies and faults in sensors in the quickest possible way, and to test their reliability and functionality upon detection and subsequent repair. We use sensor data available from industrial usage, and create an Artificial Neural Network system to carry out prognostics for the next immediate occurrence of a sensor failure, by detecting anomalies in advance in its data. In industry 4.0, Recurrent Neural Network systems utilise the fact that readings of a sensor are not arbitrary, but rather correlate to a regular pattern. Under this category, the Long Short-Term Memory technique is implemented in the process to generate maximum accuracy. General machine learning algorithms like Regression Tree and Support Vector Machine are chosen for comparison with Artificial Neural Networks to establish maximum possible efficiency and accuracy for the latter in the predictive process. These methods have many modes of application in fields of prediction and classification due to which they have been specifically chosen to perform the same tasks that the Artificial Neural Network is exposed to. Therefore, using these predictor and clustering based methods on the available datasets, the system generates a new predictive model that'll help determine the occurrence of a sensor failure beforehand.