IoT for Industrial Operations - overview
Exploratory analysis of time-series data generated from production systems in process industry poses several challenges. The equipments may be operating in different modes at different times, the control parameters may be set based on the specific goals (e.g. reduction in energy consumption, improvement in productivity, etc.) This further influences the operating states of the machines and the distribution of the data getting generated. Different type of analysis may be required depending on the category of operating conditions which are very specific to a production environment (plant-to-plant variation, equipment age specific variation, etc.). Exploratory data analysis tool that facilitates quick generation of insights on operating condition categories and associating appropriate time-series segments is essential before any model building exercise.
Data from industrial processes poses several unique challenges for the above goal. Typically the data consists of several inter-related measurements over a long time and it is highly non-stationary, due to process changes, operator experiences, extraneous factors etc. They could come from different operating states (planned/unplanned shutdown, transition/test runs etc). Moreover, some of the data fields is just observed whereas some of them is explicitly controlled by the plant operators.
We are building an interactive data exploration platform to obtain rich insights from the process data. Our platform is designed to carry out analysis that can assist domain experts as well as data modelers who do not have domain expertise.