Application of artificial neural networks in transient stability assessment and security monitoring of power systems
Two techniques for the transient stability and security monitoring of power systems have been introduced employing the artificial neural networks (ANNs). The first of these techniques for the rapid assessment of transient stability is based on the spectral analysis of speed deviations in the rotors of the machines during faulted operation. The maximum power spectral density estimate (using FFT) of excess kinetic energy has been used as the main feature for the classification of machine(s)/system stability. The fault tolerance of the methodology has also been explored. The second technique is based on the features obtained from a state estimator to estimate/calculate the values of critical clearing times to monitor the security level of a power systems in real-time for various operating environments. The concept of multi-region division of a typical power system into smaller regions is also introduced. This reduces the processing time at the ANNs without sacrificing the accuracy of the results. The back-propagation algorithm has been used to train a three layer feed-forward neural network. Two power networks have been studied in detail for the implementation of adopted methodologies for the assessment of transient stability and security monitoring of the power systems. ^
Major Professor: Gerald T. Heydt, Purdue University.
Engineering, Electronics and Electrical|Energy
Off-Campus Purdue Users:
To access this dissertation, please log in to our