Cross-domain fault diagnosis through optimal transport for a CSTR process

Fault diagnosis is a key task for developing safer control systems, especially in chemical plants. Nonetheless, acquiring good labeled fault data involves sampling from dangerous system conditions. A possible workaround to this limitation use simulation training data-driven systems. However, due modelling errors or unknown factors, may differ distribution real-world data. This setting known as cross-domain (CDFD). We optimal transport for: (i) exploring how relate the distance between (source) and (target) distributions, (ii) matching source target distributions through framework of domain adaptation (OTDA), resulting new that follows distribution. Comparisons show OTDA outperforms other CDFD methods.