Multi-output machine learning models for kinetic data evaluation : A Fischer–Tropsch synthesis case study

Predicting the impact of input process variables on chemical processes is key to assess their performance latter. Models explaining this through a mechanistic approach are rarely readily available, complex in nature and/or require long development time. With increased automation industries and availability high-throughput experimental data, data-driven machine learning models gaining popularity due simplicity reduced computational effort. In work, multi-output Fischer–Tropsch Synthesis data generated via Single-Event MicroKinetic (SEMK) model analyzed with different (ML) such as Lasso, K Nearest Neighbors (KNN), Support Vector Regression (SVR), Artificial Neural Network (ANN) regression. Temperature pressure identified dominant variables. Among considered ML models, ANN emerged superior performing one respect benchmark SEMK results. addition, validity neural network predictions verified using so-called Shap-value interpretation technique. The relative obtained Shap values, conversion follow order temperature (1x) > (0.22x) space–time (0.1x) syngas ratio (0.03x). (0.26x) remain for light olefin selectivity, but that (0.03x) becomes comparable. This work provides reference method identification suitable prediction processes.