Manufacturing organisations have been applying statistical methods in achieving operational excellence for a long time. With Industry 4.0 and IOT capable machinery, the quality of collected data has significantly improved. The amounts of data generated and collected have forced the organisations to build/improve the data strategy.
This is what we generally see in the evolution of any organisation’s data strategy.
- Organisations are increasingly doing away with manual data collection. This is also being mandated in several cases by regulators to ensure data integrity.
- Organisations do not want to depend on people to collect and maintain the data. Manufacturing Equipment is being upgraded (sometimes plugged with devices) such that the batch execution data is uploaded to the local site servers or remote repository eventually by eod (if not in real time).
- Organisations, while they maintain equipment of multiple makes in any given process, they are also focussing on common templates for data collection.
Organisations that were already collecting some data are now focussed on centralising the data collection and processing in data lakes. - Organisations are increasingly employing data engineering teams to clean, shape and store the data in data warehouses.
- Organisations also employ Business Analysts and BI experts to create dashboards that are automatically refreshed multiple times in a day to provide the Product Quality and Process Maturity Business Intelligence to the Plant Operators, Managers and OE experts.
- Some organisations are either training their SMEs (internally) or taking help from external data consultants to make sense of the available data and to investigate root causes of failures or the variations seen.
- Some organisations have started to employ third party ML tools to process Batch Quality data and provide probable causes of variations in the process in an automated fashion.
- We have also seen instances where BI managers are looking at options to employ Gen AI to the structured data as well as the unstructured data to pinpoint and highlight the changes and their causes.
While every Business Manager would want to eventually move to the last item in the above list it is entirely dependent on the current stage of their data evolution and investments the organisation would want to make in the coming days to their infrastructure and hiring/training their teams.
New Age Machine Learning and AI
While statistical methods will continue to help in decision making to pick up projects, we have found that businesses will be benefited if they choose to move to Classical and New Age Machine Learning for Investigations and use the recommendations.
Already being used:
- Multivariate Regression
- Decision Trees
Orgs can now enhance them to:
- Random Forests and Other Boosting based algorithms
- Clustering and Dimensionality Reduction Algorithms
- Time Series Analysis for Anomaly detection
We will cover results from some of the projects in the upcoming articles.
