Unified planning of wind generators and switched capacitor banks: A multiagent clustering-based distributed approach

In this paper, a multiagent clustering-based distributed approach for the optimal planning of wind-distributed generators (DGs) and switched capacitor banks (SCBs) is proposed. First, electrical distance matrices for the power systems are constructed. Additionally, a constrained optimization problem, which includes several indices and a few constraints, for the optimal clustering of distribution networks is formulated and solved. After obtaining optimal clusters, agents are assigned to the clusters, and a second multiobjective optimization problem (MOOP) for the distributed planning of wind DGs and SCBs is formulated and assigned to a head agent. The number of objective functions in the MOOP is equal to the number of agents. The objective function of an agent consists of three indices: Annual energy losses, investment costs, and voltage enhancement. Moreover, a deep neural network architecture is designed, and four independent networks are trained with six years of wind speed data for the seasonal wind speed forecasting. Two IEEE unbalanced test feeders, one with 37 nodes and the other with 123 nodes, and eight test cases are considered for simulations. The results show that losses and costs are optimized, and the voltage unbalance of the system is reduced.