Co-Evolutionary Strategy Algorithm to the Lockage Scheduling of the Three Gorges Project

The lockage co-scheduling problem of the five locks in the Three Gorges Project is studied. Itpsilas an NP-hard combinational optimization with large dimension coursed by the multi chamber sizes, multi lockage styles and large quantity of vessels. As a traffic scheduling, the operation strategy of each waterway-controller, i.e. the lock, is the key of optimization. And because of batch processing the strategy should decide when to work and how to packing vessels. We use polynomial function to formulate the strategy and optimize the strategy by the coefficients of polynomial items. A co-evolutionary strategy algorithm is proposed for these locks to learn the efficient strategy separately and collaboratively by lockage simulation. Experiments under real historical data show that the algorithm has high performance in the static scheduling in simulation and can learn the efficient strategy for dynamic situation by select suitable samples.