ABSTRACT
Since the invention of the first digital computers in the 1940s followed by a huge progress in reducing their size and cost and increasing their calculation speed, some machines demonstrate their “ability to perform complex tasks commonly associated with intelligent beings”. This is the usual definition of the Artificial Intelligence (AI). But are these machines really intelligent? Could a computer and its algorithms run a cement plant today in an autonomous way? No (not yet?), but for sure, some AI techniques demonstrate interesting potentials to help people at all levels and all along the cement manufacturing process.
Here, the focus is on machine learning algorithms that are the main components of an AI system. Some basics of machine learning tools and development steps are covered here with the objective to better understand the examples that are shown in the second part of this chapter and demonstrating that a machine learning tool development and implementation requires knowledge of both data scientists and cement manufacturing experts.
This chapter is illustrated with various examples. Some of them are intended to cover the process (from the quarry to the cement mill including the R&D laboratory) and some to present different machine learning techniques such as the image analysis (captured by drones or with the help of an electron microscope), sound analysis (for predictive maintenance and material quality), and process data analysis (to predict the clinker free-lime content or the cement quality).
The cement manufacturing process is a very wide and complex sum of processes. “AI”, through the machine learning tool abilities shown here, has a full potential to be implemented in almost every area of the cement manufacturing ecosystem, but the full potential and a real breakthrough would be achieved by looking at the entire ecosystem to optimize the balance in process performance, customer demand, and environmental aspects.
