The Challenge

The company current data landscape exists of multiple home grown solutions using obsolete technology combines with GE Smart Intelligence solutions and Oracle ERP. The systems are not integrated and data is used on a high aggregation level, that leads to loss of details needed for analytic insights which go beyond BI reporting.

Knowledge is predominantly concentrated at a few experts which creates a culture of ‘hero’s’, where each one has their ‘own data’ gathering based on experience.

The pain points can be summarized as:
• Senior management fully depends on results communicated by the experts
• Backwards history of product and process is difficult to generate
• No possibility for proactive insight
• No capturing of lessons learned and best practice
• Analytics takes very long
• Difficulty to train new employees
• Lot of data, but no contextualization
• No proactive analytics
• ‘Hero’culture

To support the data fusion the models defined in the Smart Industrial Transformation Approach allow to contextualize the data such that the relations and the semantics of the different data is easier to understand and the integration can be executed much faster.

After the data has been fused, analytics can be used to answer business questions. One of the major questions is why quality or throughput is decreasing, which is answered by root-cause analysis. In this case, the fused data is used to determine the most influential parameters for the reduction in quality or throughput by using data mining algorithms. The process engineer is then able to indicate which of the found parameters should be investigated further. The overall approach is also often called deep learning.

Other typical questions are comparison analysis between plants, production lines, or similar products. In this case the focus is on the deviation of the outcome, like quality and throughput. However, the aim is to get a deeper understanding on where the difference originates from. Since the Smart Industrial Transformation Approach is based on models and provides the context of data, process mining techniques can be applied to analyze the actual production process and compare it with the engineered production process or with the actual production process of a different plant, line or comparable product.

In the figure below, the production process of two similar products has been visualized and compared with each other. All processing steps in light blue are common in both processes, while the red and green processing steps indicate the deviations. A process engineer now can see with this picture whether the deviations are well known and understood or whether it is worth to further investigate particular processing steps. Other users can now drill down to investigate depending on their need, from active behavior of the process, till material component and individual equipment level

The Smart Industrial Transformation Approach models allow to combine the results of the two steps since the identified parameters can be related to processing steps in the process and the interesting processing steps identified by process mining can be related to continuous process parameters

In the Smart Industrial Transformation Approach models are used in a specific way:
• Scenarios are modelled from different point of views like e.g. for a manufacturing process: the physical asset view, the machine view, the operator view and the process engineer view. These four views together describe the actual process best.
• Scenarios are modelled in the most detailed way and semi-automatically derive higher level views, thus views with lesser detail. These different views with varying level of detail are required for different groups of people. A manager just wants to see the high level picture, while a process engineer wants to see all details. By deriving the higher level views we ensure that the different levels of details are consistent with each other.

Now it is possible to visualize and have interactive stakeholder involvement providing right level of information at the right time.

Stakeholders are able to visualize the ‘As-Is’ and the ‘To-Be’ situation, and do scenario play and impact analysis to manage process and application migration.