Study of intelligent optimization method and its application on system programming of large-scale regional drainage

With the continuously rapid growth of society and economy, threat and lost of flooding and waterlogging have been substantially increasing, especially in lowlands near rivers or lakes in plain polder area, where great threat exists to the security of social lives and economy. Hydraulic conservancy engineering, like river nets, floodgates, lakes, pump stations and dykes in highlands, are organically combined as a whole system to drain away the redundant water. However, optimal planning for such a complex drainage system in plain polder region is an N-P hard problem. In order to get the problem resolved, a new improved hybrid genetic algorithms based on the thoughts of experimental optimizing method, adaptive experimental genetic algorithms (AEGA), are proposed in this paper. Experimental operation techniques, such as uniform designs, normal random distribution and variables perturbation are embedded in the simple genetic algorithms (SGA) to form the so-called AEGA. Example shows that, compared with the two-phase method (TPM) of linear programming and SGA, AEGA has better ability to get the global solution, for the reason of auto-adaptability precision, individual variety of the whole feasible space and the efficiency of searching for local extreme point. With some practical values of application, AEGA is adaptable to deal with the complex water security problems in hydraulic conservancy engineering.