Based on the principles of fairness and efficiency, an index system and a projection pursuit(PP)model were constructed for water allocation in Wenshan Zhuang and Miao Autonomous Prefecture. The novel bat algorithm(NBA)was integrated into the PP model to solve the problem of the difficulty in determining the optimal projection direction. The established NBA-PP model was used for water allocation in eight counties in Wenshan. Five typical testing functions were used to validate the results of the NBA algorithm. The simulation results were compared with the results from the basic bat algorithm(BA), the artificial bee colony algorithm(ABC), the cuckoo search algorithm(CS), and the differential evolution algorithm(DE). The results show that the NBA algorithm can effectively balance the global search and local development abilities by introducing habitat selection strategy and the self-adaptive echo-Doppler effect mechanism. Compared with the DE, CS, ABC, and BA algorithms, the NBA algorithm performs better in optimization, and it shows a higher convergence rate, higher precision, better convergence stability, and better convergence reliability. The water allocation results using the NBA-PP model are more scientific and objective than using other classification methods.