Heuristic Detection of Pharma Temperature Anomalies Using IOT and Machine Learning Techniques

In pharmaceutical companies, during the manufacturing of tablets, early detection of an error occurred on the temperature basis plays a crucial role in terms of public life as well as the investments done on its manufacturing. Anomaly detection during manufacturing is a complex task. It is very difficult and also not accurate to detect the damage occurred using traditional methodologies these days. Forecasting the anomaly before it becomes massive is important, for example, the Chernobyl nuclear disaster. Mechanistic models are known to be demanding in computational manner. Hence, a model that can detect an anomaly at the earliest possible stage and also affordable in all possible ways is of a very high demand. The field of machine learning and Internet of things has received much interest from the community of scientists. Due to the ease of implementation of machine learning in various fields, it is of sheer interest to study whether an artificial neural network or some machine learning algorithms can be a best model for the detection of anomalies regarding threshold frequencies of temperature in combination with Internet of things. Implementing the incentives lay down by government and any sorts of temperature limit (Threshold frequency) within the given span of time can be easily achieved by the model proposed, wherein it collects the temperature data for every particular elapse of time and analyzes it using machine learning algorithms. In case of any anomalies detected, it immediately alerts the people by any kind of general-purpose input and output like sound, light, or message alert.