Deep Learning-Based Optimal Switch Placement in Smart Power Distribution Systems

The reliability of power distribution systems plays a key role in enhancing the service quality of customers. To reach a reliable system, switching devices such as circuit breakers, remote-controlled switches, and manual switches should be optimally installed in power distribution systems. However, due to the nonlinearity of the problem, a large number of candidate installation points, the diversity of decision variables, and the existence of tie lines, traditional mathematical-based switch optimization models may not be practical for large-scale systems. To address this challenge, this paper proposes a novel approach based on deep learning techniques to determine the optimal number and location of switching devices in real-world power distribution systems. We employ an ensemble learning method along with explainable artificial intelligence tools to build an accurate surrogate model. The proposed learning-based method can provide a fast and effective solution to the optimal switch placement problem, without requiring complicated mathematical optimization algorithms. The proposed method is tested on a modified 11 kV power distribution system connected to Bus 4 of the Roy Billiton test system (RBTS-Bus 4). Simulation results demonstrate that the proposed surrogate model outperforms traditional mathematical-based switch optimization models in terms of scalability and computational complexity.