Tree-type Irrigation Pipe Network Planning Using an Improved Bat Algorithm

Abstract. A tree-type irrigation pipe network has a substantial role in saving water. To address the pipe network planning problem, an improved model is proposed by comprehensively considering the effects of various pipe network layouts and pipe diameters on the investment cost. The model can guarantee the water supply connectivity using the concept of high-level water nodes. Meanwhile, an improved self-adaptive bat algorithm with step-controlled and mutation mechanisms is proposed herein. The performance of the proposed algorithm is examined using 12 unimodal/multimodal functions. The experimental results reveal that the improved bat algorithm can effectively avoid local optima and exhibits high solution accuracy. Moreover, planning experiments are performed on the two pipe network cases. Results reveal that the proposed pipe network planning using the improved bat algorithm has a good precision and extendibility.