Effective impurity control is essential for sustaining high-performance plasma operation in magnetic confinement fusion devices such as ITER. The Visible Spectroscopy Reference System (VSRS) is designed to measure plasma bremsstrahlung and spectral line radiation, enabling inference of key parameters such as the effective ion charge , electron density, and impurity characteristics. This work presents Bayesian models for the VSRS developed within the Minerva scientific modelling framework. A compact model enables fast inference from spectrally integrated polychromator signals, suitable for real-time plasma control. A comprehensive full-spectrum model incorporates survey spectrometer data and applies an anomaly detection method with asymmetric predictive distributions to automatically separate bremsstrahlung from spectral line radiation. Both models are validated using synthetic data generated from the Integrated Modelling and Analysis Suite database and cross-checked against Camera and Spectroscopy Emission Ray-tracer simulations. In addition, a synchrotron model is developed to assess the feasibility of detecting visible-range emission from runaway electrons, indicating that the VSRS could contribute to disruption avoidance and mitigation strategies at ITER.
Bayesian electron density inference from JET lithium beam emission spectra using Gaussian processes