Stable operation of power plants requires early detection of equipment anomalies and appropriate response decisions. This report summarizes the development of anomaly monitoring technologies used by Mitsubishi Heavy Industries, Ltd. in remote monitoring services for power plants. First, anomaly detection based on correlations among operating parameters has enabled monitoring to advance from conventional threshold-based approaches to statistical methods, allowing changes to be identified more effectively while improving monitoring efficiency. The incorporation of past cases has also enhanced cause estimation and response decision-making. More recently, MHI has been exploring highly versatile monitoring methods that use generative AI to automate various operational processes associated with anomaly monitoring, including analysis and evaluation. These developments indicate a promising direction for the further advancement of anomaly monitoring technology.
