Abstract
Early and reliable prediction of convergence to cyclic steady state (CSS) is increasingly important in periodic downstream bioprocessing, where switching and cut decisions are tuned for a repeatable cyclic regime. This work addresses time-to-CSS (TCSS) forecasting and CSS-existence classification for multicolumn countercurrent solvent gradient purification (MCSGP) systems under run-to-run feed variability. We propose a Multi-Feature k-Nearest Neighbours (MF-kNN) framework that performs long-horizon one-shot trajectory forecasting from an early run segment. CSS outcomes are inferred by reapplying a peak-based convergence rule to the predicted trajectory, while CSS existence is predicted via neighbour-label voting. The approach uses multivariate, standardised features, run-level splits, and a windowed neighbour search to reduce computation. Hyperparameters are tuned with a CSS-oriented objective function that balances trajectory fidelity, TCSS error, and misclassification penalties. On an in-silico MCSGP dataset (98 runs; ?t = 0.2 s; 6800 steps/run) with varying initial modifier concentration, MF-kNN produces accurate full-run forecasts from early data and enables operationally useful early go/no-go decisions. Across outlets, results support accurate CSS timing inference and high CSS-existence classification accuracy (up to 100% on selected outlets), indicating MF-kNN as a transparent and deployment-ready complement to cycle-to-cycle CSS monitoring and control.