Simple Solutions That Work! Issue 11

69 MAKING YOUR INSTALLATION ASUCCESS Machine Learning) by crunching large amounts of data. When Google suggests a restaurant to you with a ‘match rating in %’ obviously it’s not some guy who in an instant analyzed your past data, likes, dislikes, reviews and ratings and in real time suggested you a restaurant for a quick lunch. In this instance, AI and ML are at work and your cell phone is the IOT enabled device. The difference in today’s world and the past is; earlier intelligence and learning capabilities came along with pairs of hands and feet—now AI & ML comes without it as rental services. WHAT IS IN IT FOR A FOUNDRY A foundry probably has more variables to control to produce a good casting than variables involved in face recognition of a slogan shouting mob. If only we could deploy simple technology to acquire data from all fronts and variables to use the tremendous power of AI and ML to control processes, additions, compositions, temperatures and suggest gating related changes—things would be much simpler for the foundry manager. WHERE DO WE START To move forward with 4.0, the keystone is the almighty data. For data to be usable for any kind of inference, its integrity and reliability is of utmost importance, as we all know the GIGO rule. That is, computers are only as good as the data that is input. When you train AI on biased data, you obviously get biased outputs. Put very simply, you cannot perform deep learning, machine learning or leverage artificial intelligence on data that is either non-existent or incorrect. If you want to go ahead with AI and ML or just would like to take more informed decisions without much manual errors, the most important task is to collect correct, un-biased, reliable data that can be fed to AI and ML or analyzed to draw conclusions. In the future, this could lead foundries to run on their own. Let’s classify devices & equipment that we interact with every day from an IOT perspective: 1. Completely analogue 2. Digital with no communication interface 3. Digital with legacy communication interface 4. Digital with industrial grade communication interface (e.g., Modbus over RS485, RS232, TCP/IP) 5. IOT enabled If we want to have meaningful conclusion, we must acquire data from all fronts and variables and that can be accomplished by a small device like V-Sync and little low cost DYI modifications to completely analogue systems. All of the above equipment can be enabled to upload data to a cloud-based database or to a local server with increasingly lesser difficulty from top of the list to the bottom. V-SYNC A data acquisition and sync module can help in this regard for equipment 2 to 5. This module on one side can communicate with industrial equipment in various protocols such as Modbus, Profinet, Canbus and on other side it can communicate to a server over TCP/IP (typically called Ethernet) or GSM (typically called cellular) network. One such module can collect data from many connected equipment on the same protocol, acquire data and further synchronize it with database on server real-time. Once data is logged in a secure database hosted may be on Google Cloud, AWS or Azure. Dashboards can be configured to have real-time analytics with Qlik, Microsoft Power BI or Tableau. Configuring dashboard does involve domain knowledge and establishing causality of data is the key factor. Example of making a simple analytical connection in a green sand high pressure moulding foundry. A molding machine fitted with squeeze pressure sensor is transmitting: • Squeeze pressure for a particular mold • Running pattern number • Mold number for the pattern An automated testing apparatus such as a VCAT Mark II sitting Continued on next page