The present invention relates to a real-time economic dispatch calculation method using a distributed neural network. The method comprises the following steps: determining initial data of a system, and describing an electrical power system in the forms of a node, a branch and a parameter; according to the problem, determining an optimization objective and a constraint condition, and building a real-time economic dispatch model; setting a topological structure between nodes; constructing a neural network for each node, and setting parameters; setting an initial variable of the neural network; using the neural network for optimization; and determining whether a termination condition is met, and outputting a result when the termination condition is met, and repeating the optimization using the neural network when the termination condition is not met. Compared with the existing discretely distributed algorithms, the method provided in the present invention substantially reduces the calculation time, and compared with the existing continuously distributed algorithms, the method can take global equality constraint and global inequality constraint in economic dispatch into consideration at the same time, such that an obtained real-time economic dispatch result is more feasible.
展开▼
