Multi-stage flash (MSF) desalination is a process used to produce distilled water from brine. MSF desalination can reduce the amount of energy consumption by utilizing the condensation heat of steam to boil incoming seawater. This Study made a fault diagnosis of a MSF desalination process using a hybrid method of signed digraph (SDG) and dynamic partial least squares (DPLS). Among many methodologies used in fault diagnosis, SDG offers a simple and graphical representation of the relationship between various process variables and has been widely used. Using the system decomposition based on an SDG, the local models of each measured variable were constructed using DPLS, which has been shown to be a powerful regression technique for problems associated with using chemical process data. The values estimated by the models were compared with the measured values to diagnose the faults. The performance of the proposed model was verified by examining and comparing the results from case Studies. The proposed method demonstrated a good diagnostic capability compared with previous methods.