Data Analysis and Machine Learning Algorithms Evaluation for Bioliq AI-based Predictive Tool

• Main Author: Samuel Simbine, Augusto
• Format: Final Year Project
• Language: English
• Published: IRC 2019
• Subjects: Q Science (General)
• Online Access: http://utpedia.utp.edu.my/20895/1/Augusto%20Simbine_22341.pdf; http://utpedia.utp.edu.my/20895/

In synthesis gas generation, there is a challenge in dealing with the strong variations in the chemical and physical properties of the input materials, optimal plant settings can be heavily influenced by input material. The processes in which biomass is processed must be designed and operated much more flexibly, due to the special features and the type of biomass with highly fluctuating compositions. The mass production hasn’t initiated due to the inefficiency of the process and the dependency on experts. EDI GmbH in collaboration with Karlsruhe Institute of Technology (KIT) are working on the solution (AI-based tool for predictive process optimization for chemical plants) for the above-mentioned problem. This final year project identified relevant parameters through literature research, analysis and expert interview, and evaluated different machine learning algorithms and identified linear regression as the most applicable and efficient with its R-square of 0.8015, qualifying it to be used for the development of a hybrid model for the AI-based tool for predictive process optimization for chemical plants.