Optimal Model on Canal water Distribution Based on Dynamic Penalty Function and Genetic Algorithm

The present optimal water delivery scheduling models are based on the assumed equal design discharges of lateral canals, which are not in accordance with practical water delivery scheduling demand in most irrigation systems. In order to solve this problem, a model of lateral canals with unequal discharges and a solution method were proposed; At present, traditional fixed penalty factor have some problem, such as it is difficulty to use unified dimension and to get a higher searching precision, besides, it prematurely converge to local optimal solution. Therefore, the thought of simulated annealing was referred to design a dynamic penalty function. In the progress of genetic operation, the SGA (Simple Genetic Algorithm) adopted adaptive crossover mutation method, and compared distinct solutions of model which based on the method in this paper, Adaptive genetic algorithm (AGA) and traditional methods used in irrigation district widely respectively. Comparing with water delivery plan compiled using traditional methods, the results illustrate that using this method can get much more reasonable lateral canals water delivery time and homogeneous discharges of upper canal. AGA can adjust the genetic controlling parameters automatically on the basis of values of individual fitness and degree of population dispersion, and get a high precision solution. So it has a higher practical value in irrigation system management.