Using ML to Improve Process Control and Costs (Cloud Next '19)

This use case is from production pilot project work done at the world’s largest brewery AbinBev at their New Jersey brewery. The solution was submitted for the Gartner Supply Chain award and is now in the final 6 list of award candidates. The customer sponsor Adam Spunberg is ready to present jointly in this session as speaker. All downtime in production or processes needs to be minimized or the productivity and the profitability will decrease.Current methods involve using predetermined rules in identifying when preventive maintenance has to be done which results in either taking a downtime ahead of time or sometimes having unplanned downtimes or quality issues due to certain production elements failing ahead of schedule””. Using Machine Learning based predictive approach can help in determining when a failure will occur and thus help determine the most optimum schedule for undertaking preventive maintenance which itself results in more predictable production operation. In addition by using Machine Learning to control the various parameters of the production line and extend the life of the production line is really breakthrough innovation which is not possible in most production operations that are in place today. In this project, we will showcase how we accomplished this. We applied complex and advanced feature engineering to identify process control parameters which help predict the extension of the life of the filter.The results showed an increase of production from a norm of 600 barrels per production run to greater than 4000 barrels per run.

GCP Technology →

Watch more:
Next ’19 ML & AI Sessions here →
Next ‘19 All Sessions playlist →

Subscribe to the GCP Channel →

Speaker(s): Salil Amonkar, Adam Spunberg

Session ID: MLAI228
product:DialogFlow,Dialogflow Enterprise Edition,AI; fullname:Rachel Levy;

Duration: 39:41
Publisher: Google Cloud
You can watch this video also at the source.