A Data-Driven Approach for Fast Reliability Calculation Under a Given Investment Strategy

It's critical to guarantee the reliability of distribution networks in the system planning and operations for development and improvements of distribution networks. In this paper, a data-driven based method is proposed for a Distribution Network Company to calculate the reliability metrics under a given investment strategy quickly. In other words, the better investment strategy can be quickly pointed out among a list of strategies. Firstly, an OpenDSS-MATLAB environment platform, which can be used to simulate realistic and detailed distribution network model, is built. Then, Latin hypercube sampling was used to generate a list of investment strategies and the platform was adopted to obtain their reliability metrics. Next, Back-propagation (BP) neural network is used to obtain a relationship between investment plans and reliability metrics. Finally, in the case study, a modified IEEE 13-bus system is used to demonstrate the effectiveness of the proposed model.