نوع مقاله : مقاله پژوهشی
1 استادیار گروه مهندسی آب و سازه های هیدرولیکی، دانشکده مهندسی عمران، دانشگاه سمنان، ایران
2 استادیار، گروه مهندسی آب و سازههای هیدرولیکی، دانشکده مهندسی عمران، دانشگاه سمنان
3 دانشجوی کارشناسی ارشد مهندسی و مدیریت منابع آب، دانشکده مهندسی عمران، دانشگاه سمنان
4 دانشجوی دکترای مهندسی و مدیریت منابع آب، دانشکده مهندسی عمران، دانشگاه سمنان
عنوان مقاله [English]
Open channels have an important role in transferring water and flood control. Of course, it imposes the high cost to water supply projects. In the present study for the first time, bat algorithm has been usedfor optimal channel design. For this purpose, five different models are considered. In the first model, trapezoidal channel and manning equation are used. The second model is such as the first model, but the trapezoidal cross-section of the channel divided into two triangular segments and one rectangular segment. In the thired one, the maximum velocity is considered as a constraint. The fourth model is such as the second model, but the permissible side slope is considered as a constraint. The fifth model resembled the first model, but Froude number is considered as a constraint. In modeling, the cost of channel construction is considered as an objective function. Optimal parameters of bat algorithm are selected by using thesensitivity analysis. The coefficient of variation of ten random runs for different models is about 0.00001 to 0.00017. The results of proposed algorithm have been compared with LINGO software solutions and several common intelligent algorithms. Based on the findings, the results of bat algorithm significantly are close to the global solutions of LINGO software. The accuracy of this algorithm is also more than the other evolutionary algorithms. According to the other results, using the bat algorithm for the different evolutionary methods hascausedto 32% decreasing in cost. Among five considered models, the second model has the lowest cost. Used second model instead the first model is lead to 47% decreasing in cost. But, adding maximum velocity, permissible side slope and Froud number constraints to the second model has caused to 65%, 39%, and 55% increase in cost. Also, the proposed algorithm converges in 100to 500 iterations.