A Data-Driven Reliability Assessment Method for Composite Power Systems

A data driven approach for reliability assessment of composite power systems have been proposed in our paper. A Multi-Layer Extreme Learning Machine (MELM) is trained to graph the relations among system states versus minimum load curtailment. This approach is a computationally efficient alternative to the traditional optimal power flow (OPF) optimizations. Moreover, the pretraining process using denoising autoencoder (DAE) is utilized to improve the feature extraction and predictive performance. By conducting case studies on the RTS-79 system, the proposed methodology was evaluated. Results show that the proposed method is efficient and effective in enhancing the assessment of reliability of composite power systems.