POWER-GEN Asia 2018

Sessions Speakers Exhibitors Products Press Releases Floor Plan Buzz

Advanced Diagnostic System to Asses Futuristic Problems (Room Garuda 10B, 1st Floor)

Advanced Diagnostic system are used to monitor the condition of assets and process based on archived data which is used to compare the condition and to decide what action needs to be taken to asses the future situation, if the condition is not normal. This system provides the graphical representation to compare the real measured signals with the archived measured signals. The deviation can be easily recognized by monitoring the residuals between expected and real data in the graph. Diagnostic system can be integrated with other applications such as Power Plant DCS to select and fetch the tag or signal data directly from it. Once the tags or signals are selected for monitoring, user can configure various models for different computations and these models can be trained with archived data to provide the expected data for the real time. Trained data refers to the time period from archive data during which the assets monitored are in normal condition. Therefore, the trained data is considered as the 'normal behavior' of an asset or component. Then, if the real time data does not meet the 'normal behavior' of an asset or is out of threshold, then diagnostic system warns or raised the alarm for preliminary analysis and to mitigate the future critical situation. The raised alarm provides the navigation to move to the associated plant display to find the problem area. The time to detect the deviations is much less than other monitoring systems. The system also predicts the time after which the deviation can occur. Once the model is set for asset monitoring it can be changed taking in consideration the continuous changes which happens in asset performance and processes in power plants. Diagnostic system not only predicts for the hardware assets available in power plant but it is also a new way to calculate and validate soft sensors. The configuration also includes defining rules which automatically detect events and deviation occurred frequently as well as repetitively.