13 August 2021 in IEEE Transactions on Power Systems
This paper proposes an optimization model for the joint planning of sectionalizing switches (SSs), including remote-controlled switches (RCSs) and manual switches (MSs), and tie lines among distribution feeders. Considering a flexible installation of SSs and tie lines (i.e., deployment of SSs on both sides of a line and construction of external and internal tie lines on of a feeder), this model is conducive to developing an economic investment scheme while preserving the expected energy not supplied (EENS) reliability criterion. First, the minimum total cost of investment, maintenance and penalty for EENS is chosen as the objective function. Next, a route analysis method is introduced to analyze particular performances of different deployment schemes for RCSs, MSs, external tie lines and internal tie lines. Considering the differences between overhead lines and underground cables, the lower bound of the restoration time of each load is estimated based on optimal restoration strategies in different fault scenarios. Then, the joint planning problem is formulated as a mixed-integer linear programming (MILP) model with the aid of algebraic operations to calculate the global optimal solution. The effectiveness of the proposed model is verified for several cases in a 54-node distribution system. Numerical results indicate that the proposed planning solution merits higher economy and lower EENS as compared to the schemes offered by other prevalent methods.
1 April 2020 in Journal of Physics: Conference Series
In this paper, a reinforcement learning method is utilized to solve the steel production scheduling problem. Based on characteristics of steel production processing, the model of hybrid flow-shop scheduling problem is constructed. Then the model is attributed to a Markov Decision Process, and corresponding states, actions, reward function are put forward. When trading off the exploration and exploitation, an improved ε -greedy policy is designed. Finally, this hybrid flow-shop scheduling model based on reinforcement learning is applied to the scheduling example of steel production processing. Compared to genetic algorithm, the reinforcement learning method gets the better result.