Fault detection and diagnosis is a critical element in the power generation sector. Early faults detection ensures that correct mitigation measures can be taken, whilst false alarms should be eschewed to avoid unnecessary cost of operation, interruption and downtime. Modern power plant is equipped with thousands of sensors for monitoring, diagnosis and sensor validation application. By utilizing these features, we can use the collected operational data to develop a data-driven condition monitoring method. Intelligent Early Warning System (IEWS) represented by Artificial Neural Network (ANN), which was developed by training the network with real operational data, can be proven useful for real-time monitoring of a power plant. In this work, an integrated data preparation method was proposed. The ANN models and the hybrid artificial intelligence (AI) of ANN with Genetic Algorithm (GA), which is able to detect steam turbine trip for Malaysia Jana Manjung (MNJ) power station were developed. The AI models adopting ANN and GA were trained with real data from the MNJ station. The developed models were capable of detecting the specific trip earlier before the actual trip occurrence was detected by the existing control system. The AI model provides a good opportunity for further research and implementation of AI in the power generation industry especially in fault detection and diagnosis initiatives.
|Number of pages||15|
|Journal||Journal of Engineering Science and Technology|
|Publication status||Published - Apr 2019|
All Science Journal Classification (ASJC) codes