Big Data is one of the potential dividends of the Industrial Internet of Things IIoT, but collecting it is only the start. According to a white paper from Omron Automation, it only becomes of value when extracted and presented in the right way. Without using Industrial Data Science, important insights for transforming production data remain hidden.

Using existing production data, experts from Frost & Sullivan identified large production process savings. For example, production efficiency increasing by about 10%, operating costs reducing almost 20% and maintenance costs falling 50%.

Developing the potential of Big Data in your own production environment is not easy, but made easier using Industrial Data Science. It is more than collecting data to build a few graphs. It is about filtering out production-relevant information from the data and presenting it to the right audience in the right way.

Industrial Data Science is a new discipline.

There is no single generally valid approach that is suitable for every company. Every solution and application need data analysis and modelling to achieve the best possible result. The CRISP-DM model, (Cross-Industry Standard Process for Data Mining) offers an adaptable tool. For ease of use, Omron Automation has simplified and tailored CRISP-DM into a new four step approach for industrial data science.

As explained in their white paper, Omron uses a four-step approach for their Data Mining model:

Preparation

The preparation is the most important phase as it involves scoping the project with the participants to define the goal. It covers what information exists, what needs collecting and may involve using a data set for feasibility assessment. Finally, the production of a report to provide insights into the expected generated value and a realistic ROI.

Analysis and application development

Here, data collected spans a longer period to get a representative picture of the machine and process behaviour. The stages are data collection; data pre-processing; data analytics, and application of the findings. They allow training and validation of machine learning models and other data processing steps.

Evaluation

Utilising the application in the production environment, checks performance and business results. If the performance fails to meet expectations, earlier project phases are run again.

Service and Maintenance

Updates or wear and tear can mean production processes and machine behaviour can change over time. So regular revalidation of the solution is necessary to ensure that the solution works well and retains its value. Also, the amount of data available is also grows and can lead to the development of better models. This needs regular reviews of existing machine learning models.

Data-driven solution do not need to include machine learning models or artificial intelligence. Sometimes effective data processing and providing the right information at the right time in the right way can be enough. An illustrative example of such a data science project found in free white paper. “Data Science Services by OMRON – How to get the full value from your factory floor data”, available for download.

Developing the potential of Big Data in a production environment is not easy, but it is worth it says Omron. It relies on Industrial Data ScienceIt for filtering production-relevant information from the data for presentation as useful information.